Thanks for your code, a small question，In file fcn8s.py upscore_pool4 = self.upscore_pool4(score_pool4[:, :, 5: (5 + upscore2.size()[2]), 5: (5 + upscore2.size()[3])] + upscore2) upscore8 = self.upscore8(score_pool3[:, :, 9: (9 + upscore_pool4.size()[2]), 9: (9 + upscore_pool4.size()[3])] + upscore_pool4) how can we know the number 5，9 , i can't find it in the FCN paper ,i am new to the image segmentation

Hi @ZijunDeng , thanks for sharing. Elegant code. I have a question about the performance on VOC or cityscape of the models, such as pspnet, gcn, etc. I find your data augmentation is simple and different from that proposed in the corresponding paper. And I think it has a big impact on the segmentation result. So, how about the mean IOU of your implementations?

Hi there, I'm aware that you are updating your repo :) but I tried to use your repo as an initial point for my own implementation! I just trained a segNet model and tried to use the learned model to predict on a query image, but I came across the following error: KeyError: 'unexpected key "module.enc1.0.weight" in state_dict'

Do you know what causes this error?

First thanks for your implementing for ResNetDUC, I try to train on the dataset CamVid, but it is trained slow and seem not converge. What should I do to get a better model?

When I train the FCN8S model it shows error at the end of epoch 1 [epoch 1], [iter 8440 / 8498], [train loss 545605.78962] [epoch 1], [iter 8460 / 8498], [train loss 545628.17753] [epoch 1], [iter 8480 / 8498], [train loss 545635.63007] Traceback (most recent call last): File "train.py", line 214, in main(args) File "train.py", line 98, in main val_loss = validate(val_loader, net, criterion, optimizer, epoch, train_args, restore_transform, visualize) File "train.py", line 190, in validate writer.add_image(snapshot_name, val_visual) File "/home/lxt/anaconda3/lib/python3.6/site-packages/tensorboard/writer.py", line 221, in add_image self.file_writer.add_summary(image(tag, img_tensor)) File "/home/lxt/anaconda3/lib/python3.6/site-packages/tensorboard/summary.py", line 174, in image tensor = np.ndarray(tensor, dtype=np.float32) TypeError: 'torch.FloatTensor' object does not support indexing

This may not be relevant and I'm new to pytorch but I'm interested to load just the pyramid pooling module features from the weights provided in numpy format here. How can it be possible? Thanks

error happened while training, hope you can help me Traceback (most recent call last): File "train/cityscapes-psp_net/train2.py", line 252, in main() File "train/cityscapes-psp_net/train2.py", line 110, in main train(train_loader, net, criterion, optimizer, curr_epoch, args, val_loader, restore_transform, visualize) File "train/cityscapes-psp_net/train2.py", line 159, in train validate(val_loader, net, criterion, optimizer, curr_epoch, train_args, restore, visualize) File "train/cityscapes-psp_net/train2.py", line 182, in validate outputs,aa = net(inputs) File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 224, in call result = self.forward(*input, **kwargs) File "/home/htjiang/Desktop/pytorch-semantic-segmentation/models/psp_net.py", line 79, in forward x = self.final(x) File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 224, in call result = self.forward(*input, **kwargs) File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/container.py", line 67, in forward input = module(input) File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 224, in call result = self.forward(*input, **kwargs) File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/conv.py", line 254, in forward self.padding, self.dilation, self.groups) File "/usr/local/lib/python2.7/dist-packages/torch/nn/functional.py", line 52, in conv2d return f(input, weight, bias) RuntimeError: CUDNN_STATUS_INTERNAL_ERROR

Hi, Zijun

Thanks for your code. I notice you use upsample_bilinear instead of conv_transposed2d to interpolate in FCN implementation. What is your consideration?

Thanks,

Thank you very much for providing your implementation. I recently used your FCN but needed to make adjustments to make it work on my environment (miniconda, python 3). I share those adjustments in case someone finds them useful.

I change some code in eval_cs to fit CityScapes,the error happened as following

File "eval/eval_cs.py", line 58, in main() File "eval/eval_cs.py", line 42, in main for vi, data in enumerate(test_loader): File "/usr/local/lib/python2.7/dist-packages/torch/utils/data/dataloader.py", line 201, in next return self._process_next_batch(batch) File "/usr/local/lib/python2.7/dist-packages/torch/utils/data/dataloader.py", line 221, in _process_next_batch raise batch.exc_type(batch.exc_msg) TypeError: Traceback (most recent call last): File "/usr/local/lib/python2.7/dist-packages/torch/utils/data/dataloader.py", line 40, in _worker_loop samples = collate_fn([dataset[i] for i in batch_indices]) File "/usr/local/lib/python2.7/dist-packages/torch/utils/data/dataloader.py", line 109, in default_collate return [default_collate(samples) for samples in transposed] File "/usr/local/lib/python2.7/dist-packages/torch/utils/data/dataloader.py", line 112, in default_collate .format(type(batch[0])))) TypeError: batch must contain tensors, numbers, dicts or lists; found <class 'PIL.Image.Image'>

After downloading PASCAL VOC 2012 from http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCdevkit_18-May-2011.tar and unzipping there is no folder/path like VOCdefkit/VOC2012/ImageSets/Segmentation to put the file seg11valid.txt into.

Hello, author!About the SEGNET's upper sampling method has a little confusion, why use the transposition convolution, which does not record the index of the pooling? Hope receive your reply！

Hello, author!About the SEGNET's upper sampling method has a little confusion, why use the transposition convolution, which does not record the index of the pooling? Hope receive your reply！

hello author,

i installed tensorboardX, but your code imports summary writer from tensorboard, gives error. how can i solve this.

if i use from tensorboard import SummaryWriter ImportError: cannot import name SummaryWriter

if i use, from tensorboardX import SummaryWriter

Traceback (most recent call last): File "/home/rahul/rahul/pytorch-semantic-segmentation-master/train/voc-fcn/train.py", line 7, in from tensorboardX import SummaryWriter File "/home/rahul/anaconda3/envs/semseg/lib/python2.7/site-packages/tensorboardX/init.py", line 5, in from .torchvis import TorchVis File "/home/rahul/anaconda3/envs/semseg/lib/python2.7/site-packages/tensorboardX/torchvis.py", line 11, in from .writer import SummaryWriter File "/home/rahul/anaconda3/envs/semseg/lib/python2.7/site-packages/tensorboardX/writer.py", line 223 logdir: Optional[str] = None, ^ SyntaxError: invalid syntax

When I try to run the train.py, it reports an error like this:

• Can't pickle local object 'main..'

is there anyone know why? And how can I reslove it?

I training FCN32 with the my onw datatset. However, i realized that the loss is empty and lead to the error above. The detailed error: C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [293,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [294,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [32,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [33,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [34,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [35,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [514,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [515,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [516,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [517,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [518,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [160,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [161,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [162,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [163,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [164,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [165,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [166,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [167,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [168,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [256,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [257,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [258,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [259,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [260,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [261,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [262,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [384,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [385,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [386,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [387,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [388,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [389,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [390,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [391,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [392,0,0] Assertion t >= 0 && t < n_classes failed. C:/cb/pytorch_1000000000000/work/aten/src/THCUNN/SpatialClassNLLCriterion.cu:106: block: [3,0,0], thread: [393,0,0] Assertion t >= 0 && t < n_classes failed. Traceback (most recent call last): File "train.py", line 245, in main(args) File "train.py", line 126, in main train(train_loader, net, criterion, optimizer, epoch, train_args) File "train.py", line 149, in train loss.backward() File "C:\Users\ADMIN\anaconda3\envs\Pytorch\lib\site-packages\torch\tensor.py", line 198, in backward torch.autograd.backward(self, gradient, retain_graph, create_graph) File "C:\Users\ADMIN\anaconda3\envs\Pytorch\lib\site-packages\torch\autograd_init_.py", line 100, in backward allow_unreachable=True) # allow_unreachable flag RuntimeError: cuDNN error: CUDNN_STATUS_NOT_INITIALIZED (createCuDNNHandle at ..\aten\src\ATen\cudnn\Handle.cpp:9) (no backtrace available).