This paper introduces SAM-PT, an extension of the Segment Anything Model (SAM) for tracking and segmenting objects in dynamic videos. SAM-PT achieves strong zero-shot performance by leveraging point selection and propagation techniques. The approach is evaluated on popular video object segmentation benchmarks and the Unidentified Video Objects (UVO) benchmark. Code is available at https://github.com/SysCV/sam-pt.

https://arxiv.org/abs//2307.01197

YouTube: https://www.youtube.com/@ArxivPapers

PODCASTS:

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers