questions on VISAM: how to fill --mot_path your_data_path?

aixiaodewugege commented 1 year ago

Can you provide a more specific guide on how to reproduce your demo with VISAM?

fengxiuyaun commented 1 year ago

for example, --mot_path=/data/Dataset/mot

aixiaodewugege commented 1 year ago

Thanks for your quick reply. But I think that it failed to load the config file, even if I use your settings. Did I miss anything?

Here is my error log:

(mmedit-sam) wushuchen@wushuchen:~/projects/SAM_test/Grounded-Segment-Anything$ python grounded_sam_visam.py --meta_arch motr --dataset_file e2e_dance --with_box_refine --query_interaction_layer QIMv2 --num_queries 10 --det_db det_db_motrv2.json --use_checkpoint --mot_path ./ --resume motrv2_dancetrack.pth --sam_checkpoint sam_vit_h_4b8939.pth --video_path DanceTrack/test/dancetrack0003

/home/wushuchen/anaconda3/envs/mmedit-sam/lib/python3.8/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead. warnings.warn( /home/wushuchen/anaconda3/envs/mmedit-sam/lib/python3.8/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or None for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing weights=None. warnings.warn(msg) Training with Self-Cross Attention. None aaaaaaaaaaaaaaaaaaaaa Traceback (most recent call last): File "grounded_samvisam.py", line 253, in detr, , _ = build_model(args) File "/home/wushuchen/projects/SAM_test/Grounded-Segment-Anything/VISAM/models/init.py", line 19, in build_model return build_func(args) File "/home/wushuchen/projects/SAM_test/Grounded-Segment-Anything/VISAM/models/motr.py", line 731, in build num_frames_per_batch = max(args.sampler_lengths) TypeError: 'NoneType' object is not iterable

And my folder structed is as followed:

fengxiuyaun commented 1 year ago

Sorry, you can use this param" --meta_arch motr --dataset_file e2e_dance --with_box_refine --batch_size 1 --sample_mode random_interval --sample_interval 10 --sampler_lengths 5 --merger_dropout 0 --dropout 0 --random_drop 0.1 --fp_ratio 0.3 --query_interaction_layer QIMv2 --query_denoise 0.05 --num_queries 10 --append_crowd --det_db det_db_motrv2.json --use_checkpoint --mot_path XXX --resume motrv2_dancetrack.pth --sam_checkpoint sam_vit_h_4b8939.pth --video_path DanceTrack/test/dancetrack0003"， I will change readme

dcjniwota commented 1 year ago

I have the same problem:TypeError: 'NoneType' object is not iterable，thank you for your replay. Other question is: what's in e2e_dance dir?

aixiaodewugege commented 1 year ago

I have the same problem:TypeError: 'NoneType' object is not iterable，thank you for your replay. Other question is: what's in e2e_dance dir?

I think it is not a dir, it means only detect people.

fengxiuyaun commented 1 year ago

Yes, because we use motrv2 weight now. After we will release our code

aixiaodewugege commented 1 year ago

Yes, because we use motrv2 weight now. After we will release our code

Sorry to bother again. Do you actually use groundingDINO in grounded_sam_visam.py? I can't find your code using groundingDINO detector.

fengxiuyaun commented 1 year ago

No.

aixiaodewugege commented 1 year ago

Ok. Thanks.

Thank you for your work, it's wonderful !

dcjniwota commented 1 year ago

Here is my error new log when using the new command: All test image data In /vdb1/dataset/visam/mot/DanceTrack/test/dancetrack0003/img1: Looking forward to your reply~