# autonomousvision/unimatch

[TPAMI'23] Unifying Flow, Stereo and Depth Estimation

Repository: https://github.com/autonomousvision/unimatch
Canonical: https://ross.abutalabs.com/products/unimatch
Homepage: https://haofeixu.github.io/unimatch/
Language: Python
License: MIT
License Family: permissive
Topics: cross-attention, depth, stereo, transformer, matching, unified-model, optical-flow, correspondence
Last push: 2025-01-04T22:09:34+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 99
- inputs: {"age_days": 1398, "days_push": 606, "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 1379, forks 131 (observed 2026-08-28T04:04:33.672328+00:00)

## What it is
UniMatch is a PyTorch research library implementing a unified transformer-based model for optical flow, stereo matching, and depth estimation (TPAMI 2023). It achieves state-of-the-art results on benchmarks like Sintel, Middlebury, and Argoverse and supports cross-task transfer learning.

## Use cases
- estimate optical flow between video frames
- compute stereo disparity maps from image pairs
- predict monocular or multi-view depth
- transfer a flow-pretrained model to stereo or depth tasks
- benchmark correspondence models on Sintel, KITTI, Middlebury, or Argoverse
- use a unified matching backbone for 3D perception research

## When to choose
- you need state-of-the-art optical flow, stereo, or depth estimation in one codebase
- you want to study or exploit cross-task transfer between correspondence tasks
- you need a strong pretrained backbone for downstream 3D or reconstruction work

## When to avoid
- you need a production-ready inference service rather than research code
- you lack a GPU or PyTorch environment
- you need real-time performance on edge devices without optimization

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, machine-learning, deep-learning, autonomous-vehicles, robotics
- platform: python
- tags: optical-flow, stereo-matching, depth-estimation, transformer, correspondence, pytorch, research-code, tpami, linux, gpu

## Member repositories
- autonomousvision/unimatch (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:33.672328+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 2026-08-30T04:40:18.128978+00:00, confidence not recorded.
  - readme: https://github.com/autonomousvision/unimatch (fetched 2026-08-28T04:04:33.672328+00:00, sha d5e0dc06aec2)
  - homepage: https://haofeixu.github.io/unimatch/ (fetched 2026-08-29T11:56:20.377910+00:00, sha cc39b8b7f039)
- Data as of 2026-08-30T08:39:29.467469+00:00.