# roboflow/rf-detr

RF-DETR is a real-time object detection and segmentation model architecture developed by Roboflow, SOTA on COCO, designed for fine-tuning. [ICLR 2026]

Repository: https://github.com/roboflow/rf-detr
Canonical: https://ross.abutalabs.com/products/rf-detr
Homepage: https://rfdetr.roboflow.com
Language: Python
License: Apache-2.0
License Family: permissive
Topics: computer-vision, detr, machine-learning, object-detection, rf-detr, instance-segmentation, sota
Last push: 2026-08-25T16:04:47+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 38
- inputs: {"age_days": 532, "days_push": 8, "days_rel": 9, "gap_med": 6, "n_releases_24m": 28}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 9063, forks 1151 (observed 2026-08-28T04:10:28.181250+00:00)

## What it is
RF-DETR is a real-time transformer-based model architecture from Roboflow for object detection, instance segmentation, and keypoint detection, built on a DINOv2 backbone and state-of-the-art on COCO. It ships as a Python package with a consistent API designed for easy fine-tuning on custom datasets.

## Use cases
- detect objects in images in real time
- fine-tune a state-of-the-art object detection model on my own dataset
- run instance segmentation on images
- detect keypoints with a pretrained model
- benchmark detection accuracy and latency on COCO
- deploy a fast detection model for edge or production use

## When to choose
- you need real-time detection with SOTA accuracy-latency trade-offs
- you want to fine-tune a modern transformer detector on custom data
- you need detection, segmentation, and keypoints behind one API

## When to avoid
- you need classification-only models
- you require a non-Python or non-GPU environment
- you need the largest XL/2XL variants under a permissive license (they use PML 1.0)

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, image-processing, deep-learning
- domain: computer-vision, machine-learning, deep-learning, image-processing
- platform: python, cross-platform
- tags: object-detection, instance-segmentation, keypoint-detection, detr, transformer, fine-tuning, real-time-inference, dinov2, sota, gpu

## Member repositories
- roboflow/rf-detr (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:28.181250+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 2026-08-29T17:24:02.171966+00:00, confidence not recorded.
  - readme: https://github.com/roboflow/rf-detr (fetched 2026-08-28T04:10:28.181250+00:00, sha cc4addc5f2b3)
  - homepage: https://rfdetr.roboflow.com (fetched 2026-08-29T08:23:54.187962+00:00, sha 36c6c3c2e4f9)
- Data as of 2026-08-30T08:39:29.467469+00:00.