Zhangjq7585
commented
4 months ago 现在又有一个新的问题，如下：
Using` cuda device training.
Using 0 dataloader workers
Traceback (most recent call last):
File "validation.py", line 218, in <module>
main(args)
File "validation.py", line 131, in main
model = FasterRCNN(backbone=backbone, num_classes=parser_data.num_classes + 1)
File "/tmp/pytorch_multiFHsig_FasterRCNN/network_files/faster_rcnn_framework.py", line 267, in __init__
raise ValueError(
ValueError: backbone should contain an attribute out_channelsspecifying the number of output channels (assumed to be thesame for all the levels
博主您好，我在用MobileNetv2作为backbone训练Faster RCNN之后，进行Validation时，加载训练好的mobile-model-24.pth会报错。 Traceback (most recent call last): File "validation.py", line 215, in
main(args)
File "validation.py", line 140, in main
model.load_state_dict(weights_dict,strict=False)
File "/root/miniconda3/envs/pytorch1.12/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1604, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for FasterRCNN:
size mismatch for rpn.head.conv.weight: copying a param with shape torch.Size([1280, 1280, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 256, 3, 3]).
size mismatch for rpn.head.conv.bias: copying a param with shape torch.Size([1280]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for rpn.head.cls_logits.weight: copying a param with shape torch.Size([15, 1280, 1, 1]) from checkpoint, the shape in current model is torch.Size([3, 256, 1, 1]).
size mismatch for rpn.head.cls_logits.bias: copying a param with shape torch.Size([15]) from checkpoint, the shape in current model is torch.Size([3]).
size mismatch for rpn.head.bbox_pred.weight: copying a param with shape torch.Size([60, 1280, 1, 1]) from checkpoint, the shape in current model is torch.Size([12, 256, 1, 1]).
size mismatch for rpn.head.bbox_pred.bias: copying a param with shape torch.Size([60]) from checkpoint, the shape in current model is torch.Size([12]).
size mismatch for roi_heads.box_head.fc6.weight: copying a param with shape torch.Size([1024, 62720]) from checkpoint, the shape in current model is torch.Size([1024, 12544]).
但是我用res50fpn作为backbone训练Faster RCNN之后，Validation的时候加载训练好的res50fpn.pth就没错。