search

zhang-tao-whu / DVIS

DVIS: Decoupled Video Instance Segmentation Framework
MIT License
124 stars 7 forks source link

Model wieghts about VSPW dataset #3

Closed Wave2689 closed 9 months ago

Wave2689 commented 1 year ago

Very good idea and project! When I test the model on the VSPW dataset, I can't find model weights about it in Model Zoo. Where can I find it? Thank you!

zhang-tao-whu commented 1 year ago

We will release pre-training weights for the VSPW dataset, likely in mid-July. In the meantime, you can utilize the VSPW configs and VIPSeg pre-training weights to conduct VSS on the VSPW dataset. For example:

python train_net_video.py --config-file configs/VSPW/swin/DVIS_Offline_SwinL_720p.yaml --num-gpus 8 --eval-only MODEL.WEIGHTS DVIS_offline_vipseg_swinl.pth
Wave2689 commented 1 year ago

Thank you very much!

zhang-tao-whu commented 9 months ago

DVIS++ has been released, which provides the training weights on VSPW. Please refer to DVIS++ for details.