search

open-mmlab / mmtracking

OpenMMLab Video Perception Toolbox. It supports Video Object Detection (VID), Multiple Object Tracking (MOT), Single Object Tracking (SOT), Video Instance Segmentation (VIS) with a unified framework.
https://mmtracking.readthedocs.io/en/latest/
Apache License 2.0
3.52k stars 588 forks source link

An error occurred when change the detector in VID #490

Open dagezhuang opened 2 years ago

dagezhuang commented 2 years ago

The default configs settings could be work in the VID task(（detector:faster_rcnn_dc5） , but the same error occurred when replacing other detectors without any changes such as retinanet/faster_rcnn_fpn(in base/models:)

File "/home/ubuntu/anaconda3/envs/mmtracking/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1051, in _call_impl return forward_call(*input, *kwargs) File "/home/ubuntu/anaconda3/envs/mmtracking/lib/python3.7/site-packages/mmcv/runner/fp16_utils.py", line 109, in new_func return old_func(args, kwargs) File "/home/ubuntu/mmtracking/mmtrack/models/vid/base.py", line 194, in forward kwargs) File "/home/ubuntu/mmtracking/mmtrack/models/vid/dff.py", line 150, in forward_train x_single = flow_warp_feats(ref_x[i], flow) File "/home/ubuntu/mmtracking/mmtrack/core/motion/flow.py", line 33, in flow_warp_feats grid = grid + flow RuntimeError: The size of tensor a (10) must match the size of tensor b (9) at non-singleton dimension 2

JingweiZhang12 commented 2 years ago

We can not locate the error directly based on the information you provided. Maybe you should set breakpoints, trace the error, and post more details.