ERROR: pytorch1.0 test size mismatch

train on coco dataset:

python trainval_net.py --dataset coco --net vgg16 --bs 12 --nw 4 --lr 0.01 --lr_decay_step 1000 --cuda --mGPUs

then test:

python demo.py --net vgg16 --checksession 1 --checkepoch 8 --checkpoint 19543 --cuda --load_dir models --dataset coco

But I got the result:

load checkpoint models/vgg16/coco/faster_rcnn_1_8_19543.pth
Traceback (most recent call last):
  File "demo.py", line 195, in <module>
    fasterRCNN.load_state_dict(checkpoint['model'])
  File "/home/xuan/anaconda3/envs/pytorch1.0/lib/python3.6/site-packages/torch/nn/modules/module.py", line 769, in load_state_dict
    self.__class__.__name__, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for vgg16:
        size mismatch for RCNN_rpn.RPN_cls_score.weight: copying a param with shape torch.Size([24, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([18, 512, 1, 1]).
        size mismatch for RCNN_rpn.RPN_cls_score.bias: copying a param with shape torch.Size([24]) from checkpoint, the shape in current model is torch.Size([18]).
        size mismatch for RCNN_rpn.RPN_bbox_pred.weight: copying a param with shape torch.Size([48, 512, 1, 1]) from checkpoint, the shape in current model is torch.Size([36, 512, 1, 1]).
        size mismatch for RCNN_rpn.RPN_bbox_pred.bias: copying a param with shape torch.Size([48]) from checkpoint, the shape in current model is torch.Size([36]).
        size mismatch for RCNN_cls_score.weight: copying a param with shape torch.Size([81, 4096]) from checkpoint, the shape in current model is torch.Size([21, 4096]).
        size mismatch for RCNN_cls_score.bias: copying a param with shape torch.Size([81]) from checkpoint, the shape in current model is torch.Size([21]).
        size mismatch for RCNN_bbox_pred.weight: copying a param with shape torch.Size([324, 4096]) from checkpoint, the shape in current model is torch.Size([84, 4096]).
        size mismatch for RCNN_bbox_pred.bias: copying a param with shape torch.Size([324]) from checkpoint, the shape in current model is torch.Size([84]).

[EMCP]

I went ahead and compared the config of my training, with the config of the demo… turns out there were some configs snuck in… before the yaml file were loaded…

see this example… https://github.com/jwyang/faster-rcnn.pytorch/blob/master/test_net.py#L118

double check, and fix your config to match how you trained and you should be gtg

[Yanhong-Li]

I knew how to work it out, modify the lib/model/utils/config.py line 295 as __C.ANCHOR_SCALES = [4,8,16,32],then the demo.py could be run successfully.