Closed garrisonz closed 4 months ago
Which script are you using to generate this?
Which script are you using to generate this?
python demo/demo_automatic.py --chunk_size 4 --img_path {img_path} --amp --temporal_setting semionline --size 480 --output {mask_root}
The SAM output might be too inconsistent. Tracking similar background objects is also generally difficult.
https://github.com/hkchengrex/Tracking-Anything-with-DEVA/assets/1321005/6c8b59da-3ac1-4fcd-93b9-b3c6a9a0f7e3
Floor tracking is poor, especially in the lower right corner of the video. Any suggestion? Thanks
This is the bench scene in 3d_ovs dataset