How to run the seg model provided with our own videos? (*￣︶￣)

Thanks for your interest.

Because we haven't prepared this code, you may follow the style of the Youtube-VIS testing script here

    
           CUDA_VISIBLE_DEVICES=0 python test.py tracking --exp_id ytvis --dataset youtube_vis --dataset_version val --pre_hm  --pre_thresh 0.5 --inference --track_thresh 0.05 --load_model $model_path --gpus 0 --clip_len 2 --trades --box_nms 0.7 --save_results --seg

and modify the demo script accordingly.