pytorch-detect-to-track
pytorch-detect-to-track copied to clipboard
RFCN bbox dimension mismatch when running demo.py using pretrained model
I got this runtime error. Any idea what cause this?
thanks!
RuntimeError: While copying the parameter named RFCN_bbox_net.weight, whose dimensions in the model are torch.Size([6076, 512, 1, 1]) and whose dimensions in the checkpoint are torch.Size([196, 512, 1, 1]).
yes,i am confused too,how to deal with it?
I have got this error when run $ python demo.py --cuda --load_dir output/models --dataset imagenet_vid+imagenet_det --vid_list ILSVRC2015_test_00000000.mp4 --checksession 1 --checkepoch 7 --checkpoint 32941
error:
RuntimeError: While copying the parameter named RFCN_rpn.RPN_cls_score.weight, whose dimensions in the model are torch.Size([18, 512, 1, 1]) and whose dimensions in the checkpoint are torch.Size([24, 512, 1, 1]).
I figured it out . You need to add --cag in your command , which is a flag for class-agnostic bbox regression
thanks a lot
At 2019-02-25 14:26:20, "diwangbruce" [email protected] wrote:
I figured it out . You need to add --cag in your command , which is a flag for class-agnostic bbox regression
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.