# MasterBin-IIAU/UNINEXT

[CVPR'23] Universal Instance Perception as Object Discovery and Retrieval

Repository: https://github.com/MasterBin-IIAU/UNINEXT
Canonical: https://ross.abutalabs.com/products/uninext
Language: Python
License: MIT
License Family: permissive
Topics: instance-segmentation, object-detection, object-tracking, perception, referring-expression-comprehension, referring-expression-segmentation, unified-model, multi-object-tracking-segmentation, multiple-object-tracking, referring-video-object-segmentation, video-instance-segmentation, single-object-tracking, video-object-segmentation
Last push: 2023-07-18T13:02:16+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 90
- inputs: {"age_days": 1270, "days_push": 1142, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1278, forks 121 (observed 2026-08-28T04:04:13.335977+00:00)

## What it is
UNINEXT is the official PyTorch implementation of the CVPR 2023 paper 'Universal Instance Perception as Object Discovery and Retrieval'. It provides a unified model that handles multiple instance-level perception tasks including object detection, instance segmentation, single and multi-object tracking, video instance segmentation, and referring expression comprehension and segmentation.

## Use cases
- detect and segment object instances in images
- track multiple objects in videos
- segment video objects from referring expressions
- comprehend referring expressions to localize objects
- run video instance segmentation on benchmarks like YouTube-VIS and OVIS
- benchmark a unified model across 20 instance perception tasks

## When to choose
- you need one unified model covering detection, tracking, and segmentation tasks
- you want state-of-the-art baselines for referring video object segmentation or multi-object tracking
- you are reproducing or extending CVPR 2023 instance perception research

## When to avoid
- you need a lightweight production-ready detector rather than a research codebase
- you lack GPU resources for training large vision models
- you need tasks outside instance-level perception such as semantic segmentation or image classification

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, image-processing, video-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, deep-learning, artificial-intelligence
- platform: python
- tags: object-detection, instance-segmentation, object-tracking, video-object-segmentation, referring-expression-comprehension, unified-model, cvpr-2023, research, video, linux, gpu

## Member repositories
- MasterBin-IIAU/UNINEXT (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:13.335977+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 2026-08-30T05:02:43.853159+00:00, confidence not recorded.
  - readme: https://github.com/MasterBin-IIAU/UNINEXT (fetched 2026-08-28T04:04:13.335977+00:00, sha 22d4cfd8b46f)
- Data as of 2026-08-30T08:39:29.467469+00:00.