AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Segment Anything

Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick

ICCV 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Prompted SAM segmentation · 16,978 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
predict multiple output masks (ambiguity-aware) instead of a single mask (ablation)SAM · LVIS v1 object proposalsmask AR@1000: 54.9 -> 59.3Improved
use ambiguity-aware multi-mask output (ablation)SAM · 23 datasets single-point human studymask quality ratingImproved
train on automatic masks only instead of all three data-engine stages (ablation)SAM · 23-dataset suitemIoU: ~0.5 mIoU lowerNo clear effect
add each data engine stage cumulatively to training data (ablation)SAM · 23-dataset suitemIoUImproved
oversample manual and semi-automatic masks by 10x when training with all stages (ablation)SAM · 23-dataset suitemIoUImproved
train on 1M images (about 10% of SA-1B) instead of all 11M (ablation)SAM · 23-dataset suitemIoUNo clear effect
reduce training data to 0.1M images (ablation)SAM · 23-dataset suitemIoUHurt
scale image encoder from ViT-B to ViT-H (ablation)SAM · 23-dataset suitemIoUImproved
scale image encoder from ViT-L to ViT-H (ablation)SAM · 23-dataset suitemIoUNo clear effect
SAM vs RITM on single-point prompts (mIoU, 23 datasets)SAM · 23-dataset suitemIoU: higher on 16 of 23 datasetsImproved
use zero-shot SAM for instance segmentation instead of fully supervised ViTDet masksSAM with ViTDet boxes · COCOmask AP: 51.0 -> 46.5Hurt
add more point prompts (1 to 9) to interactive segmentationSAM vs baselines · 23-dataset suitemIoUMixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy