SysCV/sam-hq

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

4255 stars · 264 forks observed · 2026-08-28

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

HQ-SAM (Segment Anything in High Quality) upgrades Meta's Segment Anything Model with a learnable High-Quality Output Token for accurate zero-shot segmentation of objects with intricate structures. It provides pretrained checkpoints, a Python package, and integrations with Hugging Face Transformers, OpenMMLab, and annotation tools.

Use cases

• segment objects in images with high-quality masks
• zero-shot image segmentation without task-specific training
• generate fine-grained mask annotations for datasets
• segment geospatial or satellite imagery
• promptable segmentation with points or boxes
• video object segmentation pipelines

When to choose

• you need higher mask quality than vanilla SAM, especially for intricate object boundaries
• you want zero-shot promptable segmentation with pretrained checkpoints
• you need fine-grained masks for annotation or downstream multimodal models

When to avoid

• you need real-time segmentation on CPU-constrained hardware
• you only need classification or detection rather than pixel-level masks
• you want a fully trained task-specific segmentation model

Facets

library · maturity active

computer-vision image-processing machine-learning deep-learning computer-vision image-processing machine-learning deep-learning python cross-platform segment-anything sam image-segmentation zero-shot hq-sam mask-prediction promptable-segmentation neurips-2023 gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="SysCV/sam-hq")