# IDEA-Research/MaskDINO

[CVPR 2023] Official implementation of the paper "Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation"

Repository: https://github.com/IDEA-Research/MaskDINO
Canonical: https://ross.abutalabs.com/products/maskdino
Language: Python
License: Apache-2.0
License Family: permissive
Topics: instance-segmentation, object-detection, panoptic-segmentation, semantic-segmentation
Last push: 2023-12-20T07:05:03+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1549, "days_push": 987, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1557, forks 162 (observed 2026-08-28T04:05:03.347465+00:00)

## What it is
Official PyTorch implementation of Mask DINO, a unified transformer-based framework for object detection and segmentation, built on detectron2. It provides models and training code achieving state-of-the-art results on COCO instance/panoptic segmentation and ADE20K semantic segmentation.

## Use cases
- train a state-of-the-art instance segmentation model on COCO
- run panoptic segmentation on images with a transformer detector
- perform semantic segmentation on ADE20K
- reproduce CVPR 2023 Mask DINO paper results
- fine-tune a unified detection and segmentation model
- use pretrained checkpoints for object detection and segmentation inference

## When to choose
- you need top-tier transformer-based detection and segmentation accuracy
- you want a unified model handling instance, panoptic, and semantic segmentation
- your stack already uses detectron2 or detrex
- you need official paper checkpoints for research baselines

## When to avoid
- you need a lightweight production detector with fast inference on CPU
- you want actively maintained code with frequent updates
- you need open-vocabulary or text-prompted segmentation (consider OpenSeeD or SAM instead)
- you lack GPU resources for training large vision transformers

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, image-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, deep-learning, machine-learning
- platform: python
- tags: object-detection, instance-segmentation, panoptic-segmentation, semantic-segmentation, transformer, detectron2, cvpr-2023, research-code, linux, gpu

## Member repositories
- IDEA-Research/MaskDINO (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:03.347465+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 2026-08-30T04:29:55.517830+00:00, confidence not recorded.
  - readme: https://github.com/IDEA-Research/MaskDINO (fetched 2026-08-28T04:05:03.347465+00:00, sha fd8591e5c6c4)
- Data as of 2026-08-30T08:39:29.467469+00:00.