Segment Anything

Code

• github.com/facebookresearch/segment-anything
  OfficialIn paperpytorch★ 53,742
• github.com/huggingface/transformers
  pytorch★ 158,292
• github.com/IDEA-Research/Grounded-Segment-Anything
  pytorch★ 17,472
• github.com/kornia/kornia
  pytorch★ 11,125
• github.com/PaddlePaddle/PaddleSeg
  paddle★ 9,319
• github.com/geekyutao/inpaint-anything
  pytorch★ 7,606
• github.com/ux-decoder/segment-everything-everywhere-all-at-once
  pytorch★ 4,771
• github.com/syscv/sam-hq
  pytorch★ 4,201
• github.com/umass-foundation-model/3d-llm
  pytorch★ 1,187
• github.com/computational-cell-analytics/micro-sam
  pytorch★ 666

Abstract

We introduce the Segment Anything (SA) project: a new task, model, and dataset for image segmentation. Using our efficient model in a data collection loop, we built the largest segmentation dataset to date (by far), with over 1 billion masks on 11M licensed and privacy respecting images. The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks. We evaluate its capabilities on numerous tasks and find that its zero-shot performance is impressive -- often competitive with or even superior to prior fully supervised results. We are releasing the Segment Anything Model (SAM) and corresponding dataset (SA-1B) of 1B masks and 11M images at https://segment-anything.com to foster research into foundation models for computer vision.

Tasks

Benchmark Results

Reproductions