Skip to content

开源实时目标检测模型:rf-detr,在Microsoft COCO基准测试中超过60AP 体积小,可在边缘设备上运行,能同时兼顾精度和实时性 基于DETR架构结合了DINOv2,在T4 GPU上处理每张图片约6毫秒 有两个版本: RF-DETR-base:参数量29M RF-DETR-large:参数量128MRF-DETR is a real-time object detection model architecture developed by Roboflow, SOTA on COCO & designed for fine-tuning.

Python · active 2025-03-232026-08-06 (UTC)

Partial coverage11,294 / 12,108 hourly files (93%) · 2 absent upstream · 812 failed, retryable2025-03-242026-08-10 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
5.8K
Pushes
1.1K
Pull requests
262
Issues
354
Stars
2.2K
Forks
266

Activity over time

Daily event counts in the loaded window

Line chart, 502 days from 2025-03-23 to 2026-08-06. Pushes: 1,054 total, peak 44 in a day. Pull requests: 262 total, peak 17 in a day. Issues: 354 total, peak 13 in a day. Comments: 1,138 total, peak 20 in a day. Stars: 2,214 total, peak 64 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

ContributorContributionsPushesPRsComments
Borda98468472104
Copilot4086938202
isaacrob-roboflow18545151
github-actions[bot]159136230
SkalskiP10036840
pre-commit-ci[bot]696720
codecov[bot]610061
probicheaux5717157
CLAassistant400040
isaacrob370135
Abdul-Mukit340320
Matvezy241273
MichalTurek140010
LeMinhNgan13002
zhaoshuo122313027
panagiotamoraiti13029
jrdalenberg12017
yeldarby11006
nok11019
mkaic10611

Recent activity

Latest issues, pull requests and releases

  • Issue comment#1302codecov[bot]2026-08-06 17:04
    fix(weights): don't warn about the deterministic _kp_active_mask buffer as a missing weight
  • Issue comment#1183hinogi2026-07-04 06:24
    Refactor/types rfdetr evaluation
  • Issue comment#497congbrian2026-06-28 01:20
    Does RF-DETR support configuring the input resolution to 1280×720
  • Issue#1167Borda2026-06-26 19:38
    PostProcess does not clamp predicted boxes to target image dimensions
  • Issue comment#1117codecov[bot]2026-06-15 20:52
    Fix keypoint loss scaling under grad accumulation
  • Issue#1066Borda2026-06-14 06:53
    Add ExecuTorch support for RF-DETR
  • Issue comment#1066anatoly-ryabchenko2026-06-14 00:24
    Add ExecuTorch support for RF-DETR
  • Issue comment#1052anatoly-ryabchenko2026-06-14 00:22
    Roadmap for official rectangular training and inference support?
  • Issue comment#1102codecov[bot]2026-06-13 01:25
    remove deprecated API scheduled for `1.8.0`
  • Issue comment#1097codecov[bot]2026-06-10 22:16
    docs(agents): restructure AGENTS.md as minimal agentic-only reference
  • Issue comment#974Borda2026-05-27 22:46
    RF-DETR v1.6.4 Regression: 2x Slower Training, Logging Issues, and Missing GPU Memory Bar
  • Issue comment#1037omkar-3342026-05-25 23:45
    warn and disable training resize-and-crop branch when `aug_config={}`
  • Issue comment#1069omkar-3342026-05-25 23:43
    test(tflite): mask ai_edge_litert in the interpreter mock fixture
  • Issue#1068omkar-3342026-05-25 23:35
    `TestCreateInterpreter` tests fail with "Could not open 'model.tflite'" when `ai_edge_litert` is installed
  • Pull request#1064Copilot2026-05-23 21:38
  • Pull request#1064Copilot2026-05-23 21:38
  • Issue comment#1043Borda2026-05-22 05:45
    RF-DETR has 52% higer latency than D-FINE
  • Issue comment#245isaacrob2026-05-20 21:01
    Soft-RFDETR : An optionnal Mixture of Experts Module to RFDETR
  • Issue comment#1035omkar-3342026-05-19 11:46
    Question: tflite support
  • Issue comment#688bc-arl2026-05-17 22:34
    Training Restart Issue
  • Issue comment#1043isaacrob2026-05-16 04:48
    RF-DETR has 52% higer latency than D-FINE
  • Issue#157isaacrob2026-05-13 22:03
    Let's benchmark D-FINE
  • Issue comment#1026codecov[bot]2026-05-13 11:39
    test: add regression test for RFDETRSegNano PE interpolation at custom resolution
  • Issue#1029Aasim-ComputerVision2026-05-06 05:33
    PE not getting updated at runtime when resolution is changed in v1.6.5
  • Issue#529Borda2026-05-04 09:28
    training RF-DETR from scratch

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 2,214 stars here means stars gained during the window, not the repo's star count.