SAM 2: Segment Anything in Images and Videos
Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junting Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao-yuan Wu, Ross Girshick, Piotr Dollár, Christoph Feichtenhofer
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/facebookresearch/sam2OfficialIn paperpytorch★ 18,753
- github.com/facebookresearch/segment-anythingpytorch★ 53,742
- github.com/yangchris11/samuraipytorch★ 7,049
- github.com/idea-research/grounded-sam-2pytorch★ 3,350
- github.com/bowang-lab/medsam2pytorch★ 582
- github.com/ibaiGorordo/ONNX-SAM2-Segment-Anythingnone★ 287
- github.com/TripleJoy/SAM2MOTpytorch★ 163
- github.com/louisfinner/him2sampytorch★ 25
- github.com/MindCode-4/code-4/tree/main/sammindspore★ 0
- github.com/MindSpore-scientific/code-7/tree/main/SAMnone★ 0
Abstract
We present Segment Anything Model 2 (SAM 2), a foundation model towards solving promptable visual segmentation in images and videos. We build a data engine, which improves model and data via user interaction, to collect the largest video segmentation dataset to date. Our model is a simple transformer architecture with streaming memory for real-time video processing. SAM 2 trained on our data provides strong performance across a wide range of tasks. In video segmentation, we observe better accuracy, using 3x fewer interactions than prior approaches. In image segmentation, our model is more accurate and 6x faster than the Segment Anything Model (SAM). We believe that our data, model, and insights will serve as a significant milestone for video segmentation and related perception tasks. We are releasing our main model, dataset, as well as code for model training and our demo.
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| DAVIS 2017 (val) | SAM2 | J&F | 90.7 | — | Unverified |
| MOSE | SAM2 | J&F | 77.9 | — | Unverified |