jovanavidenovic/DAM4SAM
A distractor-aware memory module that plugs into SAM2 to improve visual object tracking robustness.
Not currently ranked — collecting fresh signals.
star history
This project presents DAM4SAM, a visual object tracking system that augments the Segment Anything Model 2 (SAM2) with a distractor-aware memory mechanism. The approach addresses a key weakness of memory-based trackers by better handling distracting objects in the scene. The method achieves state-of-the-art performance on six benchmarks including VOT/S challenges without additional training. The authors also introduce the DiDi dataset for studying distractor handling in tracking.
Frequently asked
- What is jovanavidenovic/DAM4SAM?
- A distractor-aware memory module that plugs into SAM2 to improve visual object tracking robustness.
- Is DAM4SAM open source?
- Yes — jovanavidenovic/DAM4SAM is an open-source project tracked on heatdrop.
- What language is DAM4SAM written in?
- jovanavidenovic/DAM4SAM is primarily written in Python.
- How popular is DAM4SAM?
- jovanavidenovic/DAM4SAM has 483 stars on GitHub.
- Where can I find DAM4SAM?
- jovanavidenovic/DAM4SAM is on GitHub at https://github.com/jovanavidenovic/DAM4SAM.