// ◉ Multimodal & Vision
Segment Anything (SAM)
A promptable foundation model that segments any object from a click or box.
Segment Anything（SAM） 2023 Intermediate core
IN PLAIN WORDS
A heavy image encoder runs once; a light prompt encoder and mask decoder respond to points, boxes and text in real time. Released with 1.1B masks; SAM 2 extends to video.
Used in: Annotation tools, medical imaging, video editing
01Original paper
Segment Anything
Alexander Kirillov, Eric Mintun, Nikhila Ravi, et al. · 2023
arXiv ↗ ✓ Verified · arXiv API
02Where it sits
Learn the left column first; the right column builds on this one.
Learn first
Then you can learn
- Nothing in the atlas builds on it yet
Related:Masked Autoencoders (MAE)
Left: prerequisites · right: builds on this · dashed: related. Full map →
03Learning resources
Official docs, explainers, courses and reference code. All links checked.
Reference code
04Where the course teaches it
On learning paths:Generative Models & Frontier Research
For AI / data
- Verification: verified, checked 2026-10-21.
- JSON: /api/learn/algorithms.json (id:
sam)