← All works
Research paper

Segment Anything

At a glance

SAM outlines objects in images in response to prompts such as clicks or boxes.

In one paragraph

Introduced SAM, a model that outlines objects in images from prompts such as a click or a box, without retraining for each task. Released alongside a dataset of over one billion object masks across 11 million images.

Authors 12

  1. Alexander Kirillov
  2. Eric Mintun
  3. Nikhila Ravi
  4. Hanzi Mao
  5. Chloe Rolland
  6. Laura Gustafson
  7. Tete Xiao
  8. Spencer Whitehead
  9. Alexander C. Berg
  10. Wan-Yen Lo
  11. Piotr Dollár
  12. Ross Girshick

Organizations

Credited for this work at publication.