Official implementation for the paper: Multi-label Classification with Partial Annotations using Class-aware Selective Loss

Related tags

Deep Learning PartialLabelingCSL
Overview

PWC

Multi-label Classification with Partial Annotations using Class-aware Selective Loss


Paper | Pretrained models

Official PyTorch Implementation

Emanuel Ben-Baruch, Tal Ridnik, Itamar Friedman, Avi Ben-Cohen, Nadav Zamir, Asaf Noy, Lihi Zelnik-Manor
DAMO Academy, Alibaba Group

Abstract

Large-scale multi-label classification datasets are commonly, and perhaps inevitably, partially annotated. That is, only a small subset of labels are annotated per sample. Different methods for handling the missing labels induce different properties on the model and impact its accuracy. In this work, we analyze the partial labeling problem, then propose a solution based on two key ideas. First, un-annotated labels should be treated selectively according to two probability quantities: the class distribution in the overall dataset and the specific label likelihood for a given data sample. We propose to estimate the class distribution using a dedicated temporary model, and we show its improved efficiency over a naive estimation computed using the dataset's partial annotations. Second, during the training of the target model, we emphasize the contribution of annotated labels over originally un-annotated labels by using a dedicated asymmetric loss. Experiments conducted on three partially labeled datasets, OpenImages, LVIS, and simulated-COCO, demonstrate the effectiveness of our approach. Specifically, with our novel selective approach, we achieve state-of-the-art results on OpenImages dataset. Code will be made available.

Class-aware Selective Approach

An overview of our approach is summarized in the following figure:

Loss Implementation

Our loss consists of a selective approach for adjusting the training mode for each class individualy and a partial asymmetric loss.

An implementation of the Class-aware Selective Loss (CSL) can be found here.

  • class PartialSelectiveLoss(nn.Module)

Pretrained Models

We provide models pretrained on the OpenImages datasset with different modes and architectures:

Model Architecture Link mAP
Ignore TResNet-M link 85.38
Negative TResNet-M link 85.85
Selective (CSL) TResNet-M link 86.72
Selective (CSL) TResNet-L link 87.34

Inference Code (Demo)

We provide inference code, that demonstrate how to load the model, pre-process an image and do inference. Example run of OpenImages model (after downloading the relevant model):

python infer.py  \
--dataset_type=OpenImages \
--model_name=tresnet_m \
--model_path=./models_local/mtresnet_opim_86.72.pth \
--pic_path=./pics/10162266293_c7634cbda9_o.jpg \
--input_size=448

Result Examples

Training Code

Training code is provided in (train.py). Also, code for simulating partial annotation for the MS-COCO dataset is available (here). In particular, two "partial" simulation schemes are implemented: fix-per-class(FPC) and random-per-sample (RPS).

  • FPC: For each class, we randomly sample a fixed number of positive annotations and the same number of negative annotations. The rest of the annotations are dropped.
  • RPA: We omit each annotation with probability p.

Pretrained weights using the ImageNet-21k dataset can be found here: link
Pretrained weights using the ImageNet-1k dataset can be found here: link

Example of training with RPS simulation:

--data=/mnt/datasets/COCO/COCO_2014
--model-path=models/pretrain/mtresnet_21k
--gamma_pos=0
--gamma_neg=4
--gamma_unann=4
--simulate_partial_type=rps
--simulate_partial_param=0.5
--partial_loss_mode=selective
--likelihood_topk=5
--prior_threshold=0.5
--prior_path=./outputs/priors/prior_fpc_1000.csv

Example of training with FPC simulation:

--data=/mnt/datasets/COCO/COCO_2014
--model-path=models/pretrain/mtresnet_21k
--gamma_pos=0
--gamma_neg=4
--gamma_unann=4
--simulate_partial_type=fpc
--simulate_partial_param=1000
--partial_loss_mode=selective
--likelihood_topk=5
--prior_threshold=0.5
--prior_path=./outputs/priors/prior_fpc_1000.csv

Typical Training Results

FPC (1,000) simulation scheme:

Model mAP
Ignore, CE 76.46
Negative, CE 81.24
Negative, ASL (4,1) 81.64
CSL - Selective, P-ASL(4,3,1) 83.44

RPS (0.5) simulation scheme:

Model mAP
Ignore, CE 84.90
Negative, CE 81.21
Negative, ASL (4,1) 81.91
CSL- Selective, P-ASL(4,1,1) 85.21

Estimating the Class Distribution

The training code contains also the procedure for estimting the class distribution from the data. Our approach enables to rank the classes based on training a temporary model usinig the Ignore mode. link

Top 10 classes:

Method Top 10 ranked classes
Original 'person', 'chair', 'car', 'dining table', 'cup', 'bottle', 'bowl', 'handbag', 'truck', 'backpack'
Estiimate (Ignore mode) 'person', 'chair', 'handbag', 'cup', 'bench', 'bottle', 'backpack', 'car', 'cell phone', 'potted plant'
Estimate (Negative mode) 'kite' 'truck' 'carrot' 'baseball glove' 'tennis racket' 'remote' 'cat' 'tie' 'horse' 'boat'

Citation

@misc{benbaruch2021multilabel,
      title={Multi-label Classification with Partial Annotations using Class-aware Selective Loss}, 
      author={Emanuel Ben-Baruch and Tal Ridnik and Itamar Friedman and Avi Ben-Cohen and Nadav Zamir and Asaf Noy and Lihi Zelnik-Manor},
      year={2021},
      eprint={2110.10955},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Acknowledgements

Several images from OpenImages dataset are used in this project. ֿ
Some components of this code implementation are adapted from the repository https://github.com/Alibaba-MIIL/ASL.

Comments
  • OID-V6 dataset preprocessing

    OID-V6 dataset preprocessing

    Amazing paper, especially so soon after the last one, great work! I had one question regarding the oidv6 dataset: did you do any preprocessing to filter out bad classes etc? How many of the 9 million images did you end up using? Also, are there any large differeneces between train.py (COCO) and what you used to train on the oidv6 dataset?

    Thanks in advance.

    opened by Leterax 8
  • Soft labels

    Soft labels

    The correctness (or concordance with the equations in your paper) of the one_side_w (and also asymmetric_w below) relies on the fact that your labels (or targets y) are hard labels (0 or 1). However, when one uses soft labels (using label smoothing), the concordance would fail.

    I think the following code works for both hard labels (identical) and soft labels (not considering efficiency)

        def forward(self, x, y):
        """"
        Parameters
        ----------
        x: input logits
        y: targets (multi-label binarized vector)
        """

        # Calculating Probabilities
        x_sigmoid = torch.sigmoid(x)
        xs_pos = x_sigmoid
        xs_neg = 1 - x_sigmoid

        # Asymmetric Clipping
        if self.clip is not None and self.clip > 0:
            xs_neg = (xs_neg + self.clip).clamp(max=1)

        # Basic CE calculation
        los_pos = y*torch.log(xs_pos.clamp(min=self.eps))
        los_neg = (1-y)*torch.log(xs_neg.clamp(min=self.eps))
        # loss = los_pos + los_neg

        # Asymmetric Focusing
        if self.gamma_neg > 0 or self.gamma_pos > 0:
            if self.disable_torch_grad_focal_loss:
                prev = torch.is_grad_enabled()
                torch.set_grad_enabled(False)
            los_pos *= torch.pow(1-xs_pos, self.gamma_pos)
            los_neg *= torch.pow(xs_pos, self.gamma_neg)
            if self.disable_torch_grad_focal_loss:
                torch.set_grad_enabled(prev)
        loss = los_pos + los_neg

        return -loss.sum()
    opened by wenh06 2
  • Issue while loading TResNet-M model

    Issue while loading TResNet-M model

    Hi,

    I'm trying to load mtresnet_opim_86.72.pth and an error occurs while loading state_dict: model.load_state_dict(state['model'], strict=True) error: Exception has occurred: RuntimeError Error(s) in loading state_dict for TResNet: Missing key(s) in state_dict: "head.fc.weight", "head.fc.bias". Unexpected key(s) in state_dict: "head.fc.embedding_generator.0.weight", "head.fc.embedding_generator.0.bias", "head.fc.FC.weight", "head.fc.FC.bias".

    If strict loading is disabled, then model is loaded without error, but classes are not found in the provided example image.

    Does the model depend on exact PyTorch, CUDA versions? I`m running it on Windows: PyTorch: 1.10.2 CUDA: 11.3

    Or I`m missing something else?

    opened by DMatHome 0
  • inference not running

    inference not running

    I get the following error when running infer.py

    Inference demo with CSL model Creating and loading the model... Traceback (most recent call last): File "infer.py", line 112, in <module> main() File "infer.py", line 68, in main model = create_model(args).cuda() File "/content/PartialLabelingCSL/src/models/utils/factory.py", line 11, in create_model model_params = {'args': args, 'num_classes': args.num_classes} AttributeError: 'Namespace' object has no attribute 'num_classes'

    Command to reproduce - python infer.py
    --dataset_type=OpenImages
    --model_name=tresnet_m
    --model_path=/content/drive/MyDrive/mtresnet_opim_86.72.pth
    --pic_path=/content/wildlife-painting-art-500x500.jpeg
    --input_size=224

    opened by jmayank23 2
  • Error while loading

    Error while loading "Selective (CSL) TResNet-L" model

    Runtime Error occurs while loading state_dict for provided TResNet-L model. I think provided pretrained model and TResnetL class differ. Selective (CSL) - TResNet-M works without any problems.

    Console Output

    Traceback (most recent call last): File "infer.py", line 104, in main() File "infer.py", line 70, in main model.load_state_dict(state['model'], strict=True) File "/home/user/anaconda3/envs/partial_labeling_csl/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1482, in load_state_dict raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format( RuntimeError: Error(s) in loading state_dict for TResNet: Missing key(s) in state_dict: "body.layer1.3.conv1.0.weight", "body.layer1.3.conv1.1.weight", "body.layer1.3.conv1.1.bias", "body.layer1.3.conv1.1.running_mean", "body.layer1.3.conv1.1.running_var", "body.layer1.3.conv2.0.weight", "body.layer1.3.conv2.1.weight", "body.layer1.3.conv2.1.bias", "body.layer1.3.conv2.1.running_mean", "body.layer1.3.conv2.1.running_var", "body.layer1.3.se.fc1.weight", "body.layer1.3.se.fc1.bias", "body.layer1.3.se.fc2.weight", "body.layer1.3.se.fc2.bias", "body.layer2.0.conv1.0.0.weight", "body.layer2.0.conv1.0.1.weight", "body.layer2.0.conv1.0.1.bias", "body.layer2.0.conv1.0.1.running_mean", "body.layer2.0.conv1.0.1.running_var", "body.layer2.0.conv2.0.weight", "body.layer2.0.conv2.1.weight", "body.layer2.0.conv2.1.bias", "body.layer2.0.conv2.1.running_mean", "body.layer2.0.conv2.1.running_var", "body.layer2.4.conv1.0.weight", "body.layer2.4.conv1.1.weight", "body.layer2.4.conv1.1.bias", "body.layer2.4.conv1.1.running_mean", "body.layer2.4.conv1.1.running_var", "body.layer2.4.conv2.0.weight", "body.layer2.4.conv2.1.weight", "body.layer2.4.conv2.1.bias", "body.layer2.4.conv2.1.running_mean", "body.layer2.4.conv2.1.running_var", "body.layer2.4.se.fc1.weight", "body.layer2.4.se.fc1.bias", "body.layer2.4.se.fc2.weight", "body.layer2.4.se.fc2.bias". Unexpected key(s) in state_dict: "body.layer1.0.conv3.0.weight", "body.layer1.0.conv3.1.weight", "body.layer1.0.conv3.1.bias", "body.layer1.0.conv3.1.running_mean", "body.layer1.0.conv3.1.running_var", "body.layer1.0.conv3.1.num_batches_tracked", "body.layer1.0.downsample.0.0.weight", "body.layer1.0.downsample.0.1.weight", "body.layer1.0.downsample.0.1.bias", "body.layer1.0.downsample.0.1.running_mean", "body.layer1.0.downsample.0.1.running_var", "body.layer1.0.downsample.0.1.num_batches_tracked", "body.layer1.1.conv3.0.weight", "body.layer1.1.conv3.1.weight", "body.layer1.1.conv3.1.bias", "body.layer1.1.conv3.1.running_mean", "body.layer1.1.conv3.1.running_var", "body.layer1.1.conv3.1.num_batches_tracked", "body.layer1.2.conv3.0.weight", "body.layer1.2.conv3.1.weight", "body.layer1.2.conv3.1.bias", "body.layer1.2.conv3.1.running_mean", "body.layer1.2.conv3.1.running_var", "body.layer1.2.conv3.1.num_batches_tracked", "body.layer2.0.conv3.0.weight", "body.layer2.0.conv3.1.weight", "body.layer2.0.conv3.1.bias", "body.layer2.0.conv3.1.running_mean", "body.layer2.0.conv3.1.running_var", "body.layer2.0.conv3.1.num_batches_tracked", "body.layer2.0.conv1.0.weight", "body.layer2.0.conv1.1.weight", "body.layer2.0.conv1.1.bias", "body.layer2.0.conv1.1.running_mean", "body.layer2.0.conv1.1.running_var", "body.layer2.0.conv1.1.num_batches_tracked", "body.layer2.0.conv2.0.0.weight", "body.layer2.0.conv2.0.1.weight", "body.layer2.0.conv2.0.1.bias", "body.layer2.0.conv2.0.1.running_mean", "body.layer2.0.conv2.0.1.running_var", "body.layer2.0.conv2.0.1.num_batches_tracked", "body.layer2.1.conv3.0.weight", "body.layer2.1.conv3.1.weight", "body.layer2.1.conv3.1.bias", "body.layer2.1.conv3.1.running_mean", "body.layer2.1.conv3.1.running_var", "body.layer2.1.conv3.1.num_batches_tracked", "body.layer2.2.conv3.0.weight", "body.layer2.2.conv3.1.weight", "body.layer2.2.conv3.1.bias", "body.layer2.2.conv3.1.running_mean", "body.layer2.2.conv3.1.running_var", "body.layer2.2.conv3.1.num_batches_tracked", "body.layer2.3.conv3.0.weight", "body.layer2.3.conv3.1.weight", "body.layer2.3.conv3.1.bias", "body.layer2.3.conv3.1.running_mean", "body.layer2.3.conv3.1.running_var", "body.layer2.3.conv3.1.num_batches_tracked", "body.layer3.18.conv1.0.weight", "body.layer3.18.conv1.1.weight", "body.layer3.18.conv1.1.bias", "body.layer3.18.conv1.1.running_mean", "body.layer3.18.conv1.1.running_var", "body.layer3.18.conv1.1.num_batches_tracked", "body.layer3.18.conv2.0.weight", "body.layer3.18.conv2.1.weight", "body.layer3.18.conv2.1.bias", "body.layer3.18.conv2.1.running_mean", "body.layer3.18.conv2.1.running_var", "body.layer3.18.conv2.1.num_batches_tracked", "body.layer3.18.conv3.0.weight", "body.layer3.18.conv3.1.weight", "body.layer3.18.conv3.1.bias", "body.layer3.18.conv3.1.running_mean", "body.layer3.18.conv3.1.running_var", "body.layer3.18.conv3.1.num_batches_tracked", "body.layer3.18.se.fc1.weight", "body.layer3.18.se.fc1.bias", "body.layer3.18.se.fc2.weight", "body.layer3.18.se.fc2.bias", "body.layer3.19.conv1.0.weight", "body.layer3.19.conv1.1.weight", "body.layer3.19.conv1.1.bias", "body.layer3.19.conv1.1.running_mean", "body.layer3.19.conv1.1.running_var", "body.layer3.19.conv1.1.num_batches_tracked", "body.layer3.19.conv2.0.weight", "body.layer3.19.conv2.1.weight", "body.layer3.19.conv2.1.bias", "body.layer3.19.conv2.1.running_mean", "body.layer3.19.conv2.1.running_var", "body.layer3.19.conv2.1.num_batches_tracked", "body.layer3.19.conv3.0.weight", "body.layer3.19.conv3.1.weight", "body.layer3.19.conv3.1.bias", "body.layer3.19.conv3.1.running_mean", "body.layer3.19.conv3.1.running_var", "body.layer3.19.conv3.1.num_batches_tracked", "body.layer3.19.se.fc1.weight", "body.layer3.19.se.fc1.bias", "body.layer3.19.se.fc2.weight", "body.layer3.19.se.fc2.bias", "body.layer3.20.conv1.0.weight", "body.layer3.20.conv1.1.weight", "body.layer3.20.conv1.1.bias", "body.layer3.20.conv1.1.running_mean", "body.layer3.20.conv1.1.running_var", "body.layer3.20.conv1.1.num_batches_tracked", "body.layer3.20.conv2.0.weight", "body.layer3.20.conv2.1.weight", "body.layer3.20.conv2.1.bias", "body.layer3.20.conv2.1.running_mean", "body.layer3.20.conv2.1.running_var", "body.layer3.20.conv2.1.num_batches_tracked", "body.layer3.20.conv3.0.weight", "body.layer3.20.conv3.1.weight", "body.layer3.20.conv3.1.bias", "body.layer3.20.conv3.1.running_mean", "body.layer3.20.conv3.1.running_var", "body.layer3.20.conv3.1.num_batches_tracked", "body.layer3.20.se.fc1.weight", "body.layer3.20.se.fc1.bias", "body.layer3.20.se.fc2.weight", "body.layer3.20.se.fc2.bias", "body.layer3.21.conv1.0.weight", "body.layer3.21.conv1.1.weight", "body.layer3.21.conv1.1.bias", "body.layer3.21.conv1.1.running_mean", "body.layer3.21.conv1.1.running_var", "body.layer3.21.conv1.1.num_batches_tracked", "body.layer3.21.conv2.0.weight", "body.layer3.21.conv2.1.weight", "body.layer3.21.conv2.1.bias", "body.layer3.21.conv2.1.running_mean", "body.layer3.21.conv2.1.running_var", "body.layer3.21.conv2.1.num_batches_tracked", "body.layer3.21.conv3.0.weight", "body.layer3.21.conv3.1.weight", "body.layer3.21.conv3.1.bias", "body.layer3.21.conv3.1.running_mean", "body.layer3.21.conv3.1.running_var", "body.layer3.21.conv3.1.num_batches_tracked", "body.layer3.21.se.fc1.weight", "body.layer3.21.se.fc1.bias", "body.layer3.21.se.fc2.weight", "body.layer3.21.se.fc2.bias", "body.layer3.22.conv1.0.weight", "body.layer3.22.conv1.1.weight", "body.layer3.22.conv1.1.bias", "body.layer3.22.conv1.1.running_mean", "body.layer3.22.conv1.1.running_var", "body.layer3.22.conv1.1.num_batches_tracked", "body.layer3.22.conv2.0.weight", "body.layer3.22.conv2.1.weight", "body.layer3.22.conv2.1.bias", "body.layer3.22.conv2.1.running_mean", "body.layer3.22.conv2.1.running_var", "body.layer3.22.conv2.1.num_batches_tracked", "body.layer3.22.conv3.0.weight", "body.layer3.22.conv3.1.weight", "body.layer3.22.conv3.1.bias", "body.layer3.22.conv3.1.running_mean", "body.layer3.22.conv3.1.running_var", "body.layer3.22.conv3.1.num_batches_tracked", "body.layer3.22.se.fc1.weight", "body.layer3.22.se.fc1.bias", "body.layer3.22.se.fc2.weight", "body.layer3.22.se.fc2.bias". size mismatch for body.conv1.0.weight: copying a param with shape torch.Size([64, 48, 3, 3]) from checkpoint, the shape in current model is torch.Size([76, 48, 3, 3]). size mismatch for body.conv1.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.conv1.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.conv1.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.conv1.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv1.0.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.0.conv1.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv1.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv1.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv1.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv2.0.weight: copying a param with shape torch.Size([64, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.0.conv2.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv2.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv2.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.conv2.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.0.se.fc1.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 76, 1, 1]). size mismatch for body.layer1.0.se.fc2.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 64, 1, 1]). size mismatch for body.layer1.0.se.fc2.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv1.0.weight: copying a param with shape torch.Size([64, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.1.conv1.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv1.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv1.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv1.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv2.0.weight: copying a param with shape torch.Size([64, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.1.conv2.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv2.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv2.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.conv2.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.1.se.fc1.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 76, 1, 1]). size mismatch for body.layer1.1.se.fc2.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 64, 1, 1]). size mismatch for body.layer1.1.se.fc2.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv1.0.weight: copying a param with shape torch.Size([64, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.2.conv1.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv1.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv1.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv1.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv2.0.weight: copying a param with shape torch.Size([64, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([76, 76, 3, 3]). size mismatch for body.layer1.2.conv2.1.weight: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv2.1.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv2.1.running_mean: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.conv2.1.running_var: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer1.2.se.fc1.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 76, 1, 1]). size mismatch for body.layer1.2.se.fc2.weight: copying a param with shape torch.Size([64, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([76, 64, 1, 1]). size mismatch for body.layer1.2.se.fc2.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([76]). size mismatch for body.layer2.0.downsample.1.0.weight: copying a param with shape torch.Size([512, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 76, 1, 1]). size mismatch for body.layer2.0.downsample.1.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.0.downsample.1.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.0.downsample.1.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.0.downsample.1.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.0.se.fc1.weight: copying a param with shape torch.Size([64, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 152, 1, 1]). size mismatch for body.layer2.0.se.fc2.weight: copying a param with shape torch.Size([128, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 64, 1, 1]). size mismatch for body.layer2.0.se.fc2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv1.0.weight: copying a param with shape torch.Size([128, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.1.conv1.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv1.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv1.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv1.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv2.0.weight: copying a param with shape torch.Size([128, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.1.conv2.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv2.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv2.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.conv2.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.1.se.fc1.weight: copying a param with shape torch.Size([64, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 152, 1, 1]). size mismatch for body.layer2.1.se.fc2.weight: copying a param with shape torch.Size([128, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 64, 1, 1]). size mismatch for body.layer2.1.se.fc2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv1.0.weight: copying a param with shape torch.Size([128, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.2.conv1.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv1.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv1.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv1.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv2.0.weight: copying a param with shape torch.Size([128, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.2.conv2.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv2.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv2.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.conv2.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.2.se.fc1.weight: copying a param with shape torch.Size([64, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 152, 1, 1]). size mismatch for body.layer2.2.se.fc2.weight: copying a param with shape torch.Size([128, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 64, 1, 1]). size mismatch for body.layer2.2.se.fc2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv1.0.weight: copying a param with shape torch.Size([128, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.3.conv1.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv1.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv1.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv1.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv2.0.weight: copying a param with shape torch.Size([128, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([152, 152, 3, 3]). size mismatch for body.layer2.3.conv2.1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv2.1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv2.1.running_mean: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.conv2.1.running_var: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer2.3.se.fc1.weight: copying a param with shape torch.Size([64, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([64, 152, 1, 1]). size mismatch for body.layer2.3.se.fc2.weight: copying a param with shape torch.Size([128, 64, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 64, 1, 1]). size mismatch for body.layer2.3.se.fc2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.0.conv1.0.weight: copying a param with shape torch.Size([256, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.0.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv2.0.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.0.conv2.0.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv2.0.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv2.0.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv2.0.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.0.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.0.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.downsample.1.0.weight: copying a param with shape torch.Size([1024, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 152, 1, 1]). size mismatch for body.layer3.0.downsample.1.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.downsample.1.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.downsample.1.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.downsample.1.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.0.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.0.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.0.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.0.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.1.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.1.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.1.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.1.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.1.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.1.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.1.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.1.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.1.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.1.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.1.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.2.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.2.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.2.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.2.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.2.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.2.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.2.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.2.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.2.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.2.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.2.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.3.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.3.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.3.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.3.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.3.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.3.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.3.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.3.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.3.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.3.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.3.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.4.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.4.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.4.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.4.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.4.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.4.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.4.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.4.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.4.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.4.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.4.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.5.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.5.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.5.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.5.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.5.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.5.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.5.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.5.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.5.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.5.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.5.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.6.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.6.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.6.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.6.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.6.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.6.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.6.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.6.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.6.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.6.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.6.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.7.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.7.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.7.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.7.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.7.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.7.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.7.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.7.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.7.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.7.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.7.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.8.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.8.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.8.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.8.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.8.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.8.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.8.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.8.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.8.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.8.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.8.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.9.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.9.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.9.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.9.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.9.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.9.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.9.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.9.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.9.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.9.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.9.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.10.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.10.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.10.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.10.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.10.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.10.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.10.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.10.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.10.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.10.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.10.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.11.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.11.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.11.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.11.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.11.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.11.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.11.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.11.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.11.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.11.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.11.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.12.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.12.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.12.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.12.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.12.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.12.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.12.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.12.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.12.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.12.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.12.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.13.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.13.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.13.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.13.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.13.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.13.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.13.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.13.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.13.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.13.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.13.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.14.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.14.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.14.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.14.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.14.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.14.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.14.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.14.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.14.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.14.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.14.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.15.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.15.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.15.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.15.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.15.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.15.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.15.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.15.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.15.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.15.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.15.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.16.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.16.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.16.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.16.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.16.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.16.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.16.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.16.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.16.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.16.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.16.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv1.0.weight: copying a param with shape torch.Size([256, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 1216, 1, 1]). size mismatch for body.layer3.17.conv1.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv1.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv1.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv1.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv2.0.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([304, 304, 3, 3]). size mismatch for body.layer3.17.conv2.1.weight: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv2.1.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv2.1.running_mean: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv2.1.running_var: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer3.17.conv3.0.weight: copying a param with shape torch.Size([1024, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([1216, 304, 1, 1]). size mismatch for body.layer3.17.conv3.1.weight: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.17.conv3.1.bias: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.17.conv3.1.running_mean: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.17.conv3.1.running_var: copying a param with shape torch.Size([1024]) from checkpoint, the shape in current model is torch.Size([1216]). size mismatch for body.layer3.17.se.fc1.weight: copying a param with shape torch.Size([128, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([152, 304, 1, 1]). size mismatch for body.layer3.17.se.fc1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([152]). size mismatch for body.layer3.17.se.fc2.weight: copying a param with shape torch.Size([256, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([304, 152, 1, 1]). size mismatch for body.layer3.17.se.fc2.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([304]). size mismatch for body.layer4.0.conv1.0.weight: copying a param with shape torch.Size([512, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([608, 1216, 1, 1]). size mismatch for body.layer4.0.conv1.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv1.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv1.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv1.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv2.0.0.weight: copying a param with shape torch.Size([512, 512, 3, 3]) from checkpoint, the shape in current model is torch.Size([608, 608, 3, 3]). size mismatch for body.layer4.0.conv2.0.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv2.0.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv2.0.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv2.0.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.0.conv3.0.weight: copying a param with shape torch.Size([2048, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([2432, 608, 1, 1]). size mismatch for body.layer4.0.conv3.1.weight: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.conv3.1.bias: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.conv3.1.running_mean: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.conv3.1.running_var: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.downsample.1.0.weight: copying a param with shape torch.Size([2048, 1024, 1, 1]) from checkpoint, the shape in current model is torch.Size([2432, 1216, 1, 1]). size mismatch for body.layer4.0.downsample.1.1.weight: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.downsample.1.1.bias: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.downsample.1.1.running_mean: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.0.downsample.1.1.running_var: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.1.conv1.0.weight: copying a param with shape torch.Size([512, 2048, 1, 1]) from checkpoint, the shape in current model is torch.Size([608, 2432, 1, 1]). size mismatch for body.layer4.1.conv1.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv1.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv1.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv1.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv2.0.weight: copying a param with shape torch.Size([512, 512, 3, 3]) from checkpoint, the shape in current model is torch.Size([608, 608, 3, 3]). size mismatch for body.layer4.1.conv2.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv2.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv2.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv2.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.1.conv3.0.weight: copying a param with shape torch.Size([2048, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([2432, 608, 1, 1]). size mismatch for body.layer4.1.conv3.1.weight: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.1.conv3.1.bias: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.1.conv3.1.running_mean: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.1.conv3.1.running_var: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.2.conv1.0.weight: copying a param with shape torch.Size([512, 2048, 1, 1]) from checkpoint, the shape in current model is torch.Size([608, 2432, 1, 1]). size mismatch for body.layer4.2.conv1.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv1.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv1.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv1.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv2.0.weight: copying a param with shape torch.Size([512, 512, 3, 3]) from checkpoint, the shape in current model is torch.Size([608, 608, 3, 3]). size mismatch for body.layer4.2.conv2.1.weight: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv2.1.bias: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv2.1.running_mean: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv2.1.running_var: copying a param with shape torch.Size([512]) from checkpoint, the shape in current model is torch.Size([608]). size mismatch for body.layer4.2.conv3.0.weight: copying a param with shape torch.Size([2048, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([2432, 608, 1, 1]). size mismatch for body.layer4.2.conv3.1.weight: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.2.conv3.1.bias: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.2.conv3.1.running_mean: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for body.layer4.2.conv3.1.running_var: copying a param with shape torch.Size([2048]) from checkpoint, the shape in current model is torch.Size([2432]). size mismatch for head.fc.embedding_generator.0.weight: copying a param with shape torch.Size([512, 2048]) from checkpoint, the shape in current model is torch.Size([512, 2432]).

    Environment

    OS: Ubuntu 18.04 PyTorch: 1.10.1 CUDA: 10.2

    Command to Reproduce

    python infer.py --dataset_type=OpenImages --model_name=tresnet_l --model_path=ltresnet_v2_opim_87.34.pth --pic_path=test_img.jpg

    opened by enesmsahin 2
  • Minor error fixes for performing inference

    Minor error fixes for performing inference

    • Modified image destination path since current path requires root access to call os.makedirs("/results").
    • Returned tensor_batch from inference(im, model, class_list, args) function since it is to be used in example loss calculation.
    • Removed double display of the output image.
    • Replaced torch._C.set_grad_enabled() calls with torch.set_grad_enabled() since it throws following AttributeError with PyTorch version 1.10.1:

    AttributeError: module 'torch._C' has no attribute 'set_grad_enabled'

    opened by enesmsahin 0
Owner
null
GitHub
Official implementation of our paper "LLA: Loss-aware Label Assignment for Dense Pedestrian Detection" in Pytorch.

LLA: Loss-aware Label Assignment for Dense Pedestrian Detection This project provides an implementation for "LLA: Loss-aware Label Assignment for Dens

null 35 Dec 6, 2022
Label Mask for Multi-label Classification

LM-MLC 一种基于完型填空的多标签分类算法 1 前言 本文主要介绍本人在全球人工智能技术创新大赛【赛道一】设计的一种基于完型填空(模板)的多标签分类算法：LM-MLC，该算法拟合能力很强能感知标签关联性，在多个数据集上测试表明该算法与主流算法无显著性差异，在该比赛数据集上的dev效果很好，但是由

null 52 Nov 20, 2022
Official Pytorch Implementation of: "Semantic Diversity Learning for Zero-Shot Multi-label Classification"(2021) paper

Semantic Diversity Learning for Zero-Shot Multi-label Classification Paper Official PyTorch Implementation Avi Ben-Cohen, Nadav Zamir, Emanuel Ben Bar

null 28 Aug 29, 2022
Official implementation of paper "Query2Label: A Simple Transformer Way to Multi-Label Classification".

Introdunction This is the official implementation of the paper "Query2Label: A Simple Transformer Way to Multi-Label Classification". Abstract This pa

Shilong Liu 274 Dec 28, 2022
NudeNet: Neural Nets for Nudity Classification, Detection and selective censoring

NudeNet: Neural Nets for Nudity Classification, Detection and selective censoring Uncensored version of the following image can be found at https://i.

notAI.tech 1.1k Dec 29, 2022
Official implementation for CVPR 2021 paper: Adaptive Class Suppression Loss for Long-Tail Object Detection

Adaptive Class Suppression Loss for Long-Tail Object Detection This repo is the official implementation for CVPR 2021 paper: Adaptive Class Suppressio

CASIA-IVA-Lab 67 Dec 4, 2022
Official implementation of "Open-set Label Noise Can Improve Robustness Against Inherent Label Noise" (NeurIPS 2021)

Open-set Label Noise Can Improve Robustness Against Inherent Label Noise NeurIPS 2021: This repository is the official implementation of ODNL. Require

Hongxin Wei 12 Dec 7, 2022
Simple and Robust Loss Design for Multi-Label Learning with Missing Labels

Simple and Robust Loss Design for Multi-Label Learning with Missing Labels Official PyTorch Implementation of the paper Simple and Robust Loss Design

Xinyu Huang 28 Oct 27, 2022
Official implementation of AAAI-21 paper "Label Confusion Learning to Enhance Text Classification Models"

Description: This is the official implementation of our AAAI-21 accepted paper Label Confusion Learning to Enhance Text Classification Models. The str

null 101 Nov 25, 2022
An implementation for the loss function proposed in Decoupled Contrastive Loss paper.

Decoupled-Contrastive-Learning This repository is an implementation for the loss function proposed in Decoupled Contrastive Loss paper. Requirements P

Ramin Nakhli 71 Dec 4, 2022
PyTorch implementation of Hierarchical Multi-label Text Classification: An Attention-based Recurrent Network

hierarchical-multi-label-text-classification-pytorch Hierarchical Multi-label Text Classification: An Attention-based Recurrent Network Approach This

Mingu Kang 17 Dec 13, 2022
Official pytorch implementation of "Feature Stylization and Domain-aware Contrastive Loss for Domain Generalization" ACMMM 2021 (Oral)

Feature Stylization and Domain-aware Contrastive Loss for Domain Generalization This is an official implementation of "Feature Stylization and Domain-

null 22 Sep 22, 2022
Recall Loss for Semantic Segmentation (This repo implements the paper: Recall Loss for Semantic Segmentation)

Recall Loss for Semantic Segmentation (This repo implements the paper: Recall Loss for Semantic Segmentation) Download Synthia dataset The model uses

null 32 Sep 21, 2022
Official implementation of Influence-balanced Loss for Imbalanced Visual Classification in PyTorch.

Official implementation of Influence-balanced Loss for Imbalanced Visual Classification in PyTorch.

Seulki Park 70 Jan 3, 2023
Working demo of the Multi-class and Anomaly classification model using the CLIP feature space

??️ Hindsight AI: Crime Classification With Clip About For Educational Purposes Only This is a recursive neural net trained to classify specific crime

Miles Tweed 2 Jun 5, 2022
Official code of ICCV2021 paper "Residual Attention: A Simple but Effective Method for Multi-Label Recognition"

CSRA This is the official code of ICCV 2021 paper: Residual Attention: A Simple But Effective Method for Multi-Label Recoginition Demo, Train and Vali

null 163 Dec 22, 2022
《LightXML: Transformer with dynamic negative sampling for High-Performance Extreme Multi-label Text Classification》(AAAI 2021) GitHub:

LightXML: Transformer with dynamic negative sampling for High-Performance Extreme Multi-label Text Classification

null 76 Dec 5, 2022
General Multi-label Image Classification with Transformers

General Multi-label Image Classification with Transformers Jack Lanchantin, Tianlu Wang, Vicente Ordóñez Román, Yanjun Qi Conference on Computer Visio

QData 154 Dec 21, 2022
A benchmark dataset for mesh multi-label-classification based on cube engravings introduced in MeshCNN

Double Cube Engravings This script creates a dataset for multi-label mesh clasification, with an intentionally difficult setup for point cloud classif

Yotam Erel 1 Nov 30, 2021