SAM2
An open-source video segmentation model released by Meta AI that performs frame-by-frame tracking and mask generation for arbitrary objects in video with minimal manual annotation. Users specify a target in a single frame via a click or bounding box, and the model automatically propagates the segmentation mask to preceding and subsequent frames.
Timeline (last 90 days)
Vlo uses SAM2 for video masking and sam-audio for stem separation.
SAM2 (Segment Anything Model 2) is an open-source video segmentation model released by Meta AI that can perform frame-by-frame tracking and mask generation for any object in a video with minimal manual annotation effort.
All Facts (2)
Vlo uses SAM2 for video masking and sam-audio for stem separation.
50%UnverifiedSAM2 (Segment Anything Model 2) is an open-source video segmentation model released by Meta AI that can perform frame-by-frame tracking and mask generation for any object in a video with minimal manual annotation effort.
50%