[ICCV2021] Learning to Track Objects from Unlabeled Videos

Model	VOT2016 EAO	VOT2018 EAO	VOT2020 EAO	LaSOT AUC (%)	TrackingNet AUC (%)	OTB100 AUC (%)
USOT	0.351	0.290	0.222	33.7	59.9	58.9
USOT*	0.402	0.344	0.219	35.8	61.5	57.4

Trying to train on a custom dataset, named it got10k to change as little as I can in the code and config, I ran the preprocess stated in the README and received the following flow json got10k_flow.txt.

When I try to run the train using

python -u ./scripts/train_usot.py --cfg experiments/train/USOT.yaml --gpus 0 --workers 8

I get the following error,

/content/drive/MyDrive/USOT/USOT-main
=> creating var/log/USOT
=> creating var/log/USOT/USOT_2022-08-12-13-57
Namespace(cfg='experiments/train/USOT.yaml', gpus='0', workers=8)
{'CHECKPOINT_DIR': 'var/snapshot',
 'GPUS': '0',
 'OUTPUT_DIR': 'var/log',
 'PRINT_FREQ': 10,
 'USOT': {'DATASET': {'FAR_SAMPLE': 3,
                      'GOT10K': {'ANNOTATION': '/content/drive/MyDrive/USOT/USOT-main/preprocessing/datasets_train/got10k/got10k_flow.json',
                                 'PATH': '/content/drive/MyDrive/USOT/USOT-main/dataset',
                                 'USE': 19000},
                      'LASOT': {'ANNOTATION': '/home/jlzheng/dataset/lasot_flow/train.json',
                                'PATH': '/home/jlzheng/dataset/lasot_flow/crop511/',
                                'USE': 13000},
                      'SCALE': 0.05,
                      'SCALEm': 0.18,
                      'SCALEs': 0.18,
                      'SHIFT': 4,
                      'SHIFTm': 64,
                      'SHIFTs': 64,
                      'VID': {'ANNOTATION': '/home/jlzheng/dataset/VID_flow/train.json',
                              'PATH': '/home/jlzheng/dataset/VID_flow/crop511/',
                              'USE': 14000},
                      'VIDEO_QUALITY': 0.4,
                      'YTVOS': {'ANNOTATION': '/home/jlzheng/dataset/ytvos_flow/train.json',
                                'PATH': '/home/jlzheng/dataset/ytvos_flow/crop511/',
                                'USE': 4000}},
          'TEST': {'DATA': 'GOT10K',
                   'END_EPOCH': 30,
                   'ISTRUE': False,
                   'MODEL': 'USOT',
                   'START_EPOCH': 10,
                   'THREADS': 11},
          'TRAIN': {'BASE_LR': 0.005,
                    'BATCH': 12,
                    'BATCH_STAGE_2': 12,
                    'CLS_RATIOS': [0.6, 0.5, 0.4],
                    'CLS_RATIO_SHIFT_EPOCHS': [0, 7, 10],
                    'END_EPOCH': 30,
                    'ISTRUE': True,
                    'LAMBDA_1_LIST': [0.3, 0.275, 0.25],
                    'LAMBDA_1_NAIVE': 0.2,
                    'LAMBDA_SHIFT_EPOCHS': [0, 7, 9],
                    'LAMBDA_TOTAL': 0.9,
                    'LAYERS_LR': 0.1,
                    'LR': {'KWARGS': {'end_lr': 2e-05, 'start_lr': 0.005},
                           'TYPE': 'log'},
                    'MEMORY_EPOCH': 6,
                    'MEMORY_NUM': 4,
                    'MODEL': 'USOT',
                    'MOMENTUM': 0.9,
                    'PRETRAIN': 'imagenet_pretrain.model',
                    'RESUME': False,
                    'SEARCH_SIZE': 255,
                    'START_EPOCH': 1,
                    'STRIDE': 8,
                    'TEMPLATE_SIZE': 127,
                    'TRAINABLE_LAYER': ['layer1', 'layer2', 'layer3'],
                    'UNFIX_EPOCH': 10,
                    'UNFIX_POLICY': 'log',
                    'WARMUP': {'EPOCH': 5,
                               'IFNOT': True,
                               'KWARGS': {'end_lr': 0.005,
                                          'start_lr': 0.0025,
                                          'step': 1},
                               'TYPE': 'step'},
                    'WARM_POLICY': 'step',
                    'WEIGHT_DECAY': 0.0001,
                    'WHICH_USE': ['GOT10K']}},
 'WORKERS': 8}
USOT(
  (criterion): BCEWithLogitsLoss()
  (features): ResNet50(
    (features): ResNet_plus2(
      (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), bias=False)
      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace)
      (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
      (layer1): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
          (downsample): Sequential(
            (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
            (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (2): Bottleneck(
          (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
      )
      (layer2): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), bias=False)
          (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
          (downsample): Sequential(
            (0): Conv2d(256, 512, kernel_size=(3, 3), stride=(2, 2), bias=False)
            (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (2): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (3): Bottleneck(
          (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
      )
      (layer3): Sequential(
        (0): Bottleneck(
          (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
          (downsample): Sequential(
            (0): Conv2d(512, 1024, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
            (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          )
        )
        (1): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(2, 2), dilation=(2, 2), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (2): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(2, 2), dilation=(2, 2), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (3): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(2, 2), dilation=(2, 2), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (4): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(2, 2), dilation=(2, 2), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
        (5): Bottleneck(
          (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(2, 2), dilation=(2, 2), bias=False)
          (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (relu): ReLU(inplace)
        )
      )
    )
  )
  (neck): AdjustLayer(
    (downsample): Sequential(
      (0): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    )
    (prpooling): PrRoIPool2D(kernel_size=(7, 7), spatial_scale=1.0)
  )
  (connect_model): box_tower_reg(
    (cls_encode): matrix(
      (matrix11_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix11_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix12_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(2, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix12_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(2, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix21_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(1, 2), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix21_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(1, 2), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
    )
    (reg_encode): matrix(
      (matrix11_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix11_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix12_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(2, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix12_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(2, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix21_k): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(1, 2), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (matrix21_s): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), dilation=(1, 2), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
    )
    (cls_dw): GroupDW()
    (reg_dw): GroupDW()
    (conf_fusion): Conf_Fusion(
      (conf_gen): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
      (value_gen): Sequential(
        (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
        (2): ReLU(inplace)
      )
    )
    (bbox_tower): Sequential(
      (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (2): ReLU()
      (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (4): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (5): ReLU()
      (6): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (7): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (8): ReLU()
      (9): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (10): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (11): ReLU()
    )
    (cls_tower): Sequential(
      (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (2): ReLU()
      (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (4): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (5): ReLU()
      (6): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (7): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (8): ReLU()
      (9): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (10): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (11): ReLU()
    )
    (cls_memory_tower): Sequential(
      (0): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (2): ReLU()
      (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (4): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (5): ReLU()
      (6): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (7): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (8): ReLU()
      (9): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
      (10): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (11): ReLU()
    )
    (bbox_pred): Conv2d(256, 4, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (cls_pred): Conv2d(256, 1, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (cls_memory_pred): Conv2d(256, 1, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
  )
)
load pretrained model from ./pretrain/imagenet_pretrain.model
remove prefix 'module.'
remove prefix 'feature_extractor.'
missing keys:['connect_model.reg_encode.matrix11_k.1.bias', 'connect_model.cls_encode.matrix21_k.1.running_var', 'connect_model.bbox_tower.10.bias', 'connect_model.cls_encode.matrix12_s.0.weight', 'connect_model.bbox_tower.1.running_var', 'connect_model.bbox_tower.7.running_var', 'connect_model.cls_memory_pred.bias', 'connect_model.cls_tower.7.weight', 'connect_model.bbox_tower.3.bias', 'connect_model.bias', 'connect_model.cls_memory_tower.6.weight', 'connect_model.cls_memory_tower.7.running_mean', 'connect_model.cls_encode.matrix11_s.0.weight', 'connect_model.cls_encode.matrix11_s.1.bias', 'connect_model.bbox_tower.4.bias', 'connect_model.bbox_tower.9.bias', 'connect_model.bbox_tower.4.running_var', 'connect_model.conf_fusion.value_gen.1.weight', 'connect_model.cls_encode.matrix21_k.0.weight', 'connect_model.conf_fusion.conf_gen.0.weight', 'connect_model.bbox_tower.7.running_mean', 'connect_model.bbox_tower.0.bias', 'connect_model.reg_encode.matrix21_k.0.weight', 'connect_model.cls_tower.1.running_mean', 'connect_model.reg_encode.matrix11_s.1.running_var', 'connect_model.reg_encode.matrix12_k.1.weight', 'connect_model.reg_encode.matrix11_s.1.bias', 'connect_model.bbox_tower.9.weight', 'connect_model.cls_tower.1.bias', 'connect_model.bbox_tower.6.weight', 'connect_model.cls_encode.matrix12_k.1.running_var', 'neck.downsample.1.running_var', 'connect_model.cls_encode.matrix21_k.1.bias', 'connect_model.reg_encode.matrix21_s.1.weight', 'connect_model.conf_fusion.value_gen.1.running_mean', 'connect_model.cls_encode.matrix21_s.1.bias', 'connect_model.cls_encode.matrix12_k.1.running_mean', 'connect_model.cls_tower.10.running_var', 'connect_model.cls_memory_tower.3.weight', 'connect_model.cls_memory_tower.4.weight', 'connect_model.conf_fusion.value_gen.0.weight', 'connect_model.bbox_tower.0.weight', 'connect_model.cls_memory_tower.4.running_var', 'connect_model.cls_encode.matrix12_k.1.weight', 'neck.downsample.1.bias', 'connect_model.cls_encode.matrix12_s.1.running_mean', 'connect_model.conf_fusion.conf_gen.1.running_mean', 'connect_model.cls_encode.matrix11_k.0.weight', 'connect_model.reg_encode.matrix12_k.1.running_var', 'connect_model.cls_tower.0.weight', 'connect_model.reg_dw.weight', 'connect_model.cls_encode.matrix12_k.0.weight', 'connect_model.reg_encode.matrix11_k.1.weight', 'connect_model.cls_encode.matrix21_s.1.running_var', 'connect_model.conf_fusion.value_gen.1.bias', 'connect_model.cls_tower.9.weight', 'connect_model.reg_encode.matrix12_s.1.running_mean', 'connect_model.reg_encode.matrix21_k.1.weight', 'connect_model.reg_encode.matrix11_s.1.running_mean', 'connect_model.cls_encode.matrix11_s.1.running_var', 'connect_model.cls_tower.9.bias', 'connect_model.bbox_pred.weight', 'connect_model.cls_encode.matrix11_k.1.bias', 'connect_model.cls_tower.3.bias', 'neck.downsample.0.weight', 'connect_model.bbox_tower.1.running_mean', 'connect_model.reg_encode.matrix12_k.1.bias', 'connect_model.reg_encode.matrix12_s.0.weight', 'connect_model.bbox_tower.1.weight', 'connect_model.bbox_tower.4.weight', 'connect_model.cls_tower.7.running_var', 'connect_model.conf_fusion.conf_gen.0.bias', 'connect_model.cls_encode.matrix21_s.1.weight', 'connect_model.cls_pred.bias', 'connect_model.cls_dw.weight', 'connect_model.adjust', 'connect_model.reg_encode.matrix21_k.1.bias', 'connect_model.conf_fusion.value_gen.1.running_var', 'connect_model.cls_memory_tower.1.bias', 'connect_model.reg_encode.matrix21_s.1.running_mean', 'connect_model.cls_encode.matrix21_s.0.weight', 'connect_model.conf_fusion.conf_gen.1.running_var', 'connect_model.cls_memory_tower.0.bias', 'connect_model.cls_tower.10.bias', 'connect_model.bbox_tower.10.running_var', 'connect_model.cls_memory_tower.9.bias', 'connect_model.cls_encode.matrix11_s.1.weight', 'connect_model.reg_encode.matrix12_s.1.weight', 'connect_model.reg_encode.matrix12_s.1.running_var', 'connect_model.cls_encode.matrix12_s.1.bias', 'connect_model.cls_pred.weight', 'connect_model.cls_memory_tower.7.weight', 'connect_model.cls_tower.4.weight', 'connect_model.cls_memory_tower.6.bias', 'connect_model.reg_encode.matrix11_s.0.weight', 'connect_model.bbox_tower.6.bias', 'connect_model.cls_memory_tower.4.bias', 'connect_model.cls_tower.6.bias', 'connect_model.cls_memory_tower.0.weight', 'connect_model.cls_tower.4.running_mean', 'connect_model.cls_encode.matrix21_k.1.running_mean', 'connect_model.cls_encode.matrix12_s.1.running_var', 'connect_model.cls_tower.6.weight', 'connect_model.cls_tower.4.running_var', 'connect_model.cls_memory_tower.7.bias', 'connect_model.cls_tower.1.running_var', 'connect_model.cls_tower.7.running_mean', 'neck.downsample.1.running_mean', 'connect_model.cls_tower.0.bias', 'connect_model.cls_memory_pred.weight', 'connect_model.cls_encode.matrix21_s.1.running_mean', 'connect_model.bbox_tower.4.running_mean', 'connect_model.cls_tower.10.running_mean', 'connect_model.cls_tower.10.weight', 'connect_model.conf_fusion.conf_gen.1.weight', 'connect_model.cls_tower.3.weight', 'connect_model.cls_memory_tower.10.weight', 'connect_model.cls_encode.matrix11_k.1.weight', 'connect_model.cls_memory_tower.1.running_mean', 'connect_model.bbox_tower.10.running_mean', 'connect_model.bbox_pred.bias', 'connect_model.reg_encode.matrix11_k.1.running_mean', 'connect_model.bbox_tower.7.bias', 'connect_model.cls_memory_tower.3.bias', 'connect_model.bbox_tower.1.bias', 'connect_model.reg_encode.matrix21_k.1.running_var', 'connect_model.cls_memory_tower.9.weight', 'connect_model.reg_encode.matrix12_k.1.running_mean', 'connect_model.conf_fusion.value_gen.0.bias', 'connect_model.reg_encode.matrix11_k.0.weight', 'connect_model.bbox_tower.7.weight', 'connect_model.reg_encode.matrix12_s.1.bias', 'connect_model.cls_memory_tower.10.bias', 'connect_model.reg_encode.matrix12_k.0.weight', 'connect_model.cls_encode.matrix11_k.1.running_var', 'connect_model.cls_tower.4.bias', 'connect_model.cls_encode.matrix11_k.1.running_mean', 'connect_model.reg_encode.matrix21_s.1.running_var', 'connect_model.cls_tower.1.weight', 'connect_model.cls_encode.matrix12_k.1.bias', 'neck.downsample.1.weight', 'connect_model.cls_encode.matrix11_s.1.running_mean', 'connect_model.reg_encode.matrix11_s.1.weight', 'connect_model.conf_fusion.conf_gen.1.bias', 'connect_model.reg_encode.matrix21_s.0.weight', 'connect_model.cls_memory_tower.10.running_var', 'connect_model.cls_memory_tower.1.weight', 'connect_model.reg_encode.matrix21_s.1.bias', 'connect_model.reg_encode.matrix11_k.1.running_var', 'connect_model.reg_encode.matrix21_k.1.running_mean', 'connect_model.cls_memory_tower.4.running_mean', 'connect_model.cls_memory_tower.10.running_mean', 'connect_model.cls_encode.matrix21_k.1.weight', 'connect_model.cls_tower.7.bias', 'connect_model.cls_memory_tower.1.running_var', 'connect_model.bbox_tower.3.weight', 'connect_model.bbox_tower.10.weight', 'connect_model.cls_memory_tower.7.running_var', 'connect_model.cls_encode.matrix12_s.1.weight']
unused checkpoint keys:['features.features.layer4.0.bn2.bias', 'features.features.layer4.0.bn2.weight', 'features.features.layer4.1.bn1.bias', 'features.features.layer4.2.bn2.weight', 'features.features.layer4.0.conv1.weight', 'features.features.layer4.0.bn2.running_mean', 'features.features.layer4.2.conv2.weight', 'features.features.layer4.2.bn1.running_mean', 'features.features.layer4.0.bn1.weight', 'features.features.layer4.1.bn1.weight', 'features.features.layer4.2.bn3.bias', 'features.features.layer4.2.bn3.running_mean', 'features.features.layer4.1.bn2.running_var', 'features.features.layer4.0.downsample.1.running_var', 'features.features.layer4.0.bn3.running_mean', 'features.features.layer4.0.bn3.bias', 'features.features.layer4.2.bn3.weight', 'features.features.layer4.2.bn1.running_var', 'features.features.layer4.1.conv3.weight', 'features.features.layer4.1.bn1.running_var', 'features.features.layer4.1.conv2.weight', 'features.features.layer4.1.bn2.weight', 'features.features.layer4.1.bn1.running_mean', 'features.features.layer4.1.bn3.running_var', 'features.features.layer4.0.downsample.1.bias', 'features.features.layer4.2.conv3.weight', 'features.features.layer4.2.bn2.running_var', 'features.features.layer4.0.bn3.weight', 'features.features.layer4.0.bn2.running_var', 'features.features.layer4.2.bn2.running_mean', 'features.features.layer4.1.bn2.running_mean', 'features.features.layer4.1.bn3.weight', 'features.features.layer4.0.downsample.0.weight', 'features.features.layer4.0.bn3.running_var', 'features.features.layer4.1.bn3.bias', 'features.features.layer4.1.bn2.bias', 'features.features.layer4.2.bn1.bias', 'features.features.layer4.0.conv2.weight', 'features.features.layer4.2.bn2.bias', 'features.features.layer4.0.bn1.running_mean', 'features.features.layer4.0.bn1.bias', 'features.features.layer4.0.bn1.running_var', 'features.features.layer4.2.bn1.weight', 'features.features.layer4.2.conv1.weight', 'features.features.layer4.0.conv3.weight', 'features.features.layer4.0.downsample.1.running_mean', 'features.features.layer4.1.conv1.weight', 'features.features.layer4.0.downsample.1.weight', 'features.features.layer4.2.bn3.running_var', 'features.features.layer4.1.bn3.running_mean']
==========first check trainable==========
{'params': <filter object at 0x7fab08023050>, 'lr': 0.0005}
{'params': <generator object Module.parameters at 0x7fab080517d0>, 'lr': 0.005}
{'params': <generator object Module.parameters at 0x7fab080519d0>, 'lr': 0.005}
==========double check trainable==========
trainable params:
neck.downsample.0.weight
neck.downsample.1.weight
neck.downsample.1.bias
connect_model.adjust
connect_model.bias
connect_model.cls_encode.matrix11_k.0.weight
connect_model.cls_encode.matrix11_k.1.weight
connect_model.cls_encode.matrix11_k.1.bias
connect_model.cls_encode.matrix11_s.0.weight
connect_model.cls_encode.matrix11_s.1.weight
connect_model.cls_encode.matrix11_s.1.bias
connect_model.cls_encode.matrix12_k.0.weight
connect_model.cls_encode.matrix12_k.1.weight
connect_model.cls_encode.matrix12_k.1.bias
connect_model.cls_encode.matrix12_s.0.weight
connect_model.cls_encode.matrix12_s.1.weight
connect_model.cls_encode.matrix12_s.1.bias
connect_model.cls_encode.matrix21_k.0.weight
connect_model.cls_encode.matrix21_k.1.weight
connect_model.cls_encode.matrix21_k.1.bias
connect_model.cls_encode.matrix21_s.0.weight
connect_model.cls_encode.matrix21_s.1.weight
connect_model.cls_encode.matrix21_s.1.bias
connect_model.reg_encode.matrix11_k.0.weight
connect_model.reg_encode.matrix11_k.1.weight
connect_model.reg_encode.matrix11_k.1.bias
connect_model.reg_encode.matrix11_s.0.weight
connect_model.reg_encode.matrix11_s.1.weight
connect_model.reg_encode.matrix11_s.1.bias
connect_model.reg_encode.matrix12_k.0.weight
connect_model.reg_encode.matrix12_k.1.weight
connect_model.reg_encode.matrix12_k.1.bias
connect_model.reg_encode.matrix12_s.0.weight
connect_model.reg_encode.matrix12_s.1.weight
connect_model.reg_encode.matrix12_s.1.bias
connect_model.reg_encode.matrix21_k.0.weight
connect_model.reg_encode.matrix21_k.1.weight
connect_model.reg_encode.matrix21_k.1.bias
connect_model.reg_encode.matrix21_s.0.weight
connect_model.reg_encode.matrix21_s.1.weight
connect_model.reg_encode.matrix21_s.1.bias
connect_model.cls_dw.weight
connect_model.reg_dw.weight
connect_model.conf_fusion.conf_gen.0.weight
connect_model.conf_fusion.conf_gen.0.bias
connect_model.conf_fusion.conf_gen.1.weight
connect_model.conf_fusion.conf_gen.1.bias
connect_model.conf_fusion.value_gen.0.weight
connect_model.conf_fusion.value_gen.0.bias
connect_model.conf_fusion.value_gen.1.weight
connect_model.conf_fusion.value_gen.1.bias
connect_model.bbox_tower.0.weight
connect_model.bbox_tower.0.bias
connect_model.bbox_tower.1.weight
connect_model.bbox_tower.1.bias
connect_model.bbox_tower.3.weight
connect_model.bbox_tower.3.bias
connect_model.bbox_tower.4.weight
connect_model.bbox_tower.4.bias
connect_model.bbox_tower.6.weight
connect_model.bbox_tower.6.bias
connect_model.bbox_tower.7.weight
connect_model.bbox_tower.7.bias
connect_model.bbox_tower.9.weight
connect_model.bbox_tower.9.bias
connect_model.bbox_tower.10.weight
connect_model.bbox_tower.10.bias
connect_model.cls_tower.0.weight
connect_model.cls_tower.0.bias
connect_model.cls_tower.1.weight
connect_model.cls_tower.1.bias
connect_model.cls_tower.3.weight
connect_model.cls_tower.3.bias
connect_model.cls_tower.4.weight
connect_model.cls_tower.4.bias
connect_model.cls_tower.6.weight
connect_model.cls_tower.6.bias
connect_model.cls_tower.7.weight
connect_model.cls_tower.7.bias
connect_model.cls_tower.9.weight
connect_model.cls_tower.9.bias
connect_model.cls_tower.10.weight
connect_model.cls_tower.10.bias
connect_model.cls_memory_tower.0.weight
connect_model.cls_memory_tower.0.bias
connect_model.cls_memory_tower.1.weight
connect_model.cls_memory_tower.1.bias
connect_model.cls_memory_tower.3.weight
connect_model.cls_memory_tower.3.bias
connect_model.cls_memory_tower.4.weight
connect_model.cls_memory_tower.4.bias
connect_model.cls_memory_tower.6.weight
connect_model.cls_memory_tower.6.bias
connect_model.cls_memory_tower.7.weight
connect_model.cls_memory_tower.7.bias
connect_model.cls_memory_tower.9.weight
connect_model.cls_memory_tower.9.bias
connect_model.cls_memory_tower.10.weight
connect_model.cls_memory_tower.10.bias
connect_model.bbox_pred.weight
connect_model.bbox_pred.bias
connect_model.cls_pred.weight
connect_model.cls_pred.bias
connect_model.cls_memory_pred.weight
connect_model.cls_memory_pred.bias
GPU NUM:  1
(WarmUPScheduler) lr spaces: 
[2.50000000e-03 2.87174589e-03 3.29876978e-03 3.78929142e-03
 4.35275282e-03 5.00000000e-03 3.97242620e-03 3.15603398e-03
 2.50742241e-03 1.99211009e-03 1.58270207e-03 1.25743343e-03
 9.99012299e-04 7.93700526e-04 6.30583352e-04 5.00989166e-04
 3.98028497e-04 3.16227766e-04 2.51238292e-04 1.99605115e-04
 1.58583318e-04 1.25992105e-04 1.00098868e-04 7.95270729e-05
 6.31830855e-05 5.01980288e-05 3.98815930e-05 3.16853369e-05
 2.51735325e-05 2.00000000e-05]
model prepare done
train datas: ['GOT10K']
############################################################
<class 'list'>
############################################################
Traceback (most recent call last):
  File "./scripts/train_usot.py", line 367, in <module>
    main()
  File "./scripts/train_usot.py", line 328, in main
    train_set = USOTDataset(config)
  File "/content/drive/MyDrive/USOT/USOT-main/lib/dataset_loader/datasets_usot.py", line 105, in __init__
    self.video_quality, self.far_sample)
  File "/content/drive/MyDrive/USOT/USOT-main/lib/dataset_loader/datasets_usot.py", line 472, in __init__
    self._clean()
  File "/content/drive/MyDrive/USOT/USOT-main/lib/dataset_loader/datasets_usot.py", line 498, in _clean
    frames = self.labels[video]
TypeError: list indices must be integers or slices, not list

Using google collab

about the score_map

When I only use offline siamese module to train this model for a while, I find that the cls_score stop update and the score_map visualized like below: Do you know how can I fix this? Thank you very much!

opened by LiShenglana 5
I found difficult to reproduce

my environment is pytorch1.8.1, cuda11.1, python3.8.0, batch size is 12 I can not reproduce the result using full data(VID, GOT10k, LASOT, YTVOS) where A R EAO I reproduce is 0.574 0.393 0.300. can it be reproduced using pytorch1.7?

opened by chenrxi 5
"Ninja is required to load C++ extensions"

Using /home/cscv/.cache/torch_extensions as PyTorch extensions root... Creating extension directory /home/cscv/.cache/torch_extensions/_prroi_pooling... Traceback (most recent call last): File "/home/cscv/Documents/lsl/USOT/scripts/train_usot.py", line 421, in main() File "/home/cscv/Documents/lsl/USOT/scripts/train_usot.py", line 411, in main curLR, config, writer_dict, logger, device=device) File "/home/cscv/Documents/lsl/USOT/scripts/train_usot.py", line 247, in usot_train search_memory_ir=search_memory_ir, search_bbox=search_bbox, cls_ratio=cls_ratio) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 159, in forward return self.module(*inputs[0], **kwargs[0]) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/cscv/Documents/lsl/USOT/lib/models/models.py", line 287, in forward _, zf = self.neck(zf, crop=True, pr_pool=self.pr_pool, bbox=template_bbox) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/cscv/Documents/lsl/USOT/lib/models/connect.py", line 349, in forward xf_pr = self.prpooling(x_ori, bbox) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/cscv/Documents/lsl/USOT/lib/models/prroi_pool/prroi_pool.py", line 28, in forward return prroi_pool2d(features, rois, self.pooled_height, self.pooled_width, self.spatial_scale) File "/home/cscv/Documents/lsl/USOT/lib/models/prroi_pool/functional.py", line 44, in forward _prroi_pooling = _import_prroi_pooling() File "/home/cscv/Documents/lsl/USOT/lib/models/prroi_pool/functional.py", line 33, in _import_prroi_pooling verbose=True File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 997, in load keep_intermediates=keep_intermediates) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1202, in _jit_compile with_cuda=with_cuda) File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1268, in _write_ninja_file_and_build_library verify_ninja_availability() File "/home/cscv/anaconda3/envs/usot-model2/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1323, in verify_ninja_availability raise RuntimeError("Ninja is required to load C++ extensions") RuntimeError: Ninja is required to load C++ extensions

I encounterd this problem, it seems like need ninja, but I already install this packages: (usot-model2) cscv@cscv-202206:~/Documents/lsl/USOT/scripts$ which ninja /home/cscv/anaconda3/envs/usot-model2/bin/ninja How can I solve this, thank you very much!

opened by LiShenglana 4
VOT2020 test

Hello, how did you complete the VOT2020 test evaluation? I have seen your comment that it is done by using test_vot2020.py file and vot toolkit. Could you please tell me more about it? Thank you again for your excellent work.

opened by zrz1018 4
Problem when increasing batchsize

When increasing the Batchsize from the default 12 to 64 this error occurs: File "/USOT/lib/models/models.py", line 276, in forward bbox_pred_to_img = self.pred_offset_to_image_bbox(off_forward_bbox, batch) File "/USOT/lib/models/models.py", line 145, in pred_offset_to_image_bbox pred_x1 = self.grid_to_search_x - bbox_pred[:, 0, ...].unsqueeze(1) RuntimeError: The size of tensor a (64) must match the size of tensor b (256) at non-singleton dimension 0

The sizes of the relevant variables for batch size 12: self.grid_to_search_x / y = [48, 1, 25, 25] bbox_pred = [48, 4, 25, 25]

and for batch size 64: self.grid_to_search_x / y = [64, 1, 25, 25] bbox_pred = [256, 4, 25, 25]

opened by wr0112358 2

error while trying to train

opened by nadavmisgav 1

Data set related issues

The got10_flow.json file I downloaded directly from Google cloud drive only reached 9334 in the end, but the GOT10k training data set has 9335 files, may I ask what is the situation?

opened by zrz1018 1
why detach(stop-gradient) the boxes for PrPooling

The contribution of precise ROI-pooling can be concluded as: pooling more precise ROI features and can backpropagate the gradients to the coordinates.

We find you detach the boxes when pooling the results in the training pipeline at https://github.com/VISION-SJTU/USOT/blob/main/lib/models/models.py#L273. that's somewhat confusing for using cycle consistency to train an unsupervised tracking model. If dont detach these boxes, the model may acquire more info from the video-level training and can be better trained ? But now this part is detached, the model seems like the conventional pair-wise trained siamese tracking model. So can we ask, why this part is detached ? what if this part is not detached?

opened by FlorinShum 1

[ICCV2021] Learning to Track Objects from Unlabeled Videos

Related tags

Overview

Unsupervised Single Object Tracking (USOT)

Introduction

Results

Tutorial

Environments

Preparations

Test and Eval

Train

Citation

Reference

Contact

Comments

Owner

The implementation of the algorithm in the paper "Safe Deep Semi-Supervised Learning for Unseen-Class Unlabeled Data" published in ICML 2020.

[ICML 2021] Break-It-Fix-It: Learning to Repair Programs from Unlabeled Data

The NEOSSat is a dual-mission microsatellite designed to detect potentially hazardous Earth-orbit-crossing asteroids and track objects that reside in deep space

Ranking Models in Unlabeled New Environments （iccv21）

A DeepStack custom model for detecting common objects in dark/night images and videos.

Search and filter videos based on objects that appear in them using convolutional neural networks

Parametric Contrastive Learning (ICCV2021)

Semi-Supervised Learning, Object Detection, ICCV2021

Dynamic Attentive Graph Learning for Image Restoration, ICCV2021 [PyTorch Code]

Official Repo for ICCV2021 Paper: Learning to Regress Bodies from Images using Differentiable Semantic Rendering

PyKale is a PyTorch library for multimodal learning and transfer learning as well as deep learning and dimensionality reduction on graphs, images, texts, and videos

Official PyTorch Implementation of Rank & Sort Loss [ICCV2021]

source code of “Visual Saliency Transformer” (ICCV2021)

HiFT: Hierarchical Feature Transformer for Aerial Tracking (ICCV2021)

Implementation of ICCV2021(Oral) paper - VMNet: Voxel-Mesh Network for Geodesic-aware 3D Semantic Segmentation

Implementation for our ICCV2021 paper: Internal Video Inpainting by Implicit Long-range Propagation

This is the pytorch implementation for the paper: Generalizable Mixed-Precision Quantization via Attribution Rank Preservation, which is accepted to ICCV2021.

Official code for "Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformer. ICCV2021".

Seeing Dynamic Scene in the Dark: High-Quality Video Dataset with Mechatronic Alignment (ICCV2021)