# UX-Decoder/Semantic-SAM

[ECCV 2024] Official implementation of the paper "Semantic-SAM: Segment and Recognize Anything at Any Granularity"

Repository: https://github.com/UX-Decoder/Semantic-SAM
Canonical: https://ross.abutalabs.com/products/semantic-sam
Language: Python
License Family: other
Last push: 2025-07-10T08:06:44+00:00

## Health v2 (maintenance only)
Score: 33/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 31, release rhythm 8, longevity 82
- inputs: {"age_days": 1150, "days_push": 419, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2854, forks 145 (observed 2026-08-28T04:07:25.786366+00:00)

## What it is
Official PyTorch implementation of Semantic-SAM, a universal image segmentation model that segments and recognizes anything at any desired granularity, from semantic to instance to part level. It reproduces SAM training on SA-1B and extends it with controllable multi-granularity interactive and automatic mask generation.

## Use cases
- segment objects in images at multiple granularity levels
- interactive click-based segmentation with multiple mask options
- auto-generate all masks for an image with controllable granularity
- reproduce SAM training on the SA-1B dataset
- part-level and semantic-aware segmentation
- research on universal image segmentation models

## When to choose
- you need segmentation masks at semantic, instance, or part level from a single model
- you want SAM-like segmentation with finer granularity control
- you need training code and checkpoints for SAM-style models
- you are doing research on interactive or universal segmentation

## When to avoid
- you need a production-ready product with a polished UI rather than research code
- you lack GPU resources for large vision transformer inference or training
- you need a permissively licensed dependency and cannot accept a missing license
- you only need simple off-the-shelf object detection without segmentation

## Facets
- artifact type: library
- maturity: active
- function: computer-vision, image-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, artificial-intelligence, deep-learning
- platform: python
- tags: image-segmentation, segment-anything, sam, interactive-segmentation, multi-granularity, eccv-2024, research-code, linux, gpu

## Member repositories
- UX-Decoder/Semantic-SAM (main) score 33

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:25.786366+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:48:30.689862+00:00, confidence not recorded.
  - readme: https://github.com/UX-Decoder/Semantic-SAM (fetched 2026-08-28T04:07:25.786366+00:00, sha 439c452e98f5)
- Data as of 2026-08-30T08:39:29.467469+00:00.