CASIA-LMC-Lab/FastSAM

Health v2 · maintenance only

Adoption not part of the score

8401 stars · 760 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

FastSAM is a CNN-based Segment Anything Model trained on only 2% of the SA-1B dataset, achieving comparable segmentation performance to SAM at 50x faster run-time speed. It supports promptable segmentation via text, points, and boxes, and is integrated into the Ultralytics YOLOv8 model hub.

Use cases

• segment all objects in an image quickly
• fast alternative to Meta's Segment Anything Model
• segment objects with point or box prompts
• text-prompted image segmentation
• real-time instance segmentation on GPU
• generate segmentation masks for downstream vision tasks

When to choose

• you need SAM-like segmentation but with much lower latency
• you want a lightweight CNN segmentation model deployable with TensorRT or Ultralytics
• you need promptable segmentation (points, boxes, text) in a Python pipeline

When to avoid

• you need the highest possible segmentation accuracy regardless of speed
• you require a transformer-based SAM with the full SA-1B training
• your project cannot use AGPL-3.0 licensed code

Facets

library · maturity active

machine-learning computer-vision image-processing deep-learning computer-vision image-processing machine-learning deep-learning python cross-platform segment-anything image-segmentation yolov8 fastsam promptable-segmentation gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="CASIA-LMC-Lab/FastSAM")