# ashkamath/mdetr

Repository: https://github.com/ashkamath/mdetr
Canonical: https://ross.abutalabs.com/products/mdetr
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2022-10-03T19:35:46+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 100
- inputs: {"age_days": 1959, "days_push": 1430, "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 1052, forks 138 (observed 2026-08-28T04:03:23.553223+00:00)

## What it is
MDETR (Modulated Detection) is a PyTorch research codebase for end-to-end multi-modal object detection that grounds free-form text queries to image regions. It includes pre-trained weights and fine-tuning scripts for phrase grounding, referring expression comprehension/segmentation, and visual question answering.

## Use cases
- detect objects mentioned in a free-form text query within an image
- phrase grounding on Flickr30k
- referring expression comprehension on RefCOCO datasets
- referring expression segmentation with PhraseCut
- pre-train a vision-language detector on aligned image-text data
- fine-tune a detector for visual question answering tasks

## When to choose
- you need text-conditioned object detection or phrase grounding with pre-trained models
- you are reproducing or building on the MDETR paper for multi-modal understanding research
- you want open-vocabulary detection where class labels come from the text query

## When to avoid
- you need a production-ready, actively maintained detection library
- you only need standard class-labeled object detection without text queries
- you lack GPU resources, since training requires multi-node distributed setups

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, nlp
- domain: computer-vision, machine-learning, artificial-intelligence
- platform: python
- tags: multimodal, object-detection, phrase-grounding, referring-expression-comprehension, vision-language, research-code, pytorch, natural-language-processing, linux, gpu

## Member repositories
- ashkamath/mdetr (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:23.553223+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 2026-08-30T06:59:41.709242+00:00, confidence not recorded.
  - readme: https://github.com/ashkamath/mdetr (fetched 2026-08-28T04:03:23.553223+00:00, sha 7e865287f949)
- Data as of 2026-08-30T08:39:29.467469+00:00.