Welcome to the official community tracker for the RT-DETR (Real-Time DEtection TRansformer) family! This repository is dedicated to curating the exciting advancements and follow-up works based on the original RT-DETR, the first real-time end-to-end detector based on the Transformer architecture.

RT-DETR is a groundbreaking object detection model that, for the first time, brings real-time performance to DETR.

Traditionally, DETR (DEtection TRansformer) models were known for their high accuracy and elegant, non-NMS (Non-Maximum Suppression) pipeline, but they suffered from slow inference speeds. RT-DETR successfully bridges this gap, establishing itself as the first real-time end-to-end Transformer detector, providing a powerful and efficient solution that rivals the performance of top-tier YOLO models.

The core architecture of RT-DETR features a hybrid encoder and a query selection mechanism, enabling it to process multi-scale features efficiently while maintaining the end-to-end simplicity of DETR.

Since the release of the original paper, the research community has enthusiastically embraced and built upon the RT-DETR framework. Thanks to these collective efforts, the RT-DETR family is rapidly growing, with numerous works enhancing its speed, accuracy, and capabilities.

This repository aims to track these valuable contributions. The evolving family includes:

DETRs Beat YOLOs on Real-time Object Detection (The foundational RT-DETR paper) [CVPR 2024]
Yian Zhao^* , Wenyu Lv^* , Shangliang Xu, Jinman Wei, Guanzhong Wang, Qingqing Dang, Yi Liu, Jie Chen

RT-DETRv2: Improved Baseline with Bag-of-Freebies for Real-Time Detection Transformer
Wenyu Lv, Yian Zhao, Qinyao Chang, Kui Huang, Guanzhong Wang, Yi Liu

RT-DETRv3: Real-time End-to-End Object Detection with Hierarchical Dense Positive Supervision [WACV 2025 Oral]
Shuo Wang^* , Chunlong Xia^* , Feng Lv, Yifeng Shi

D-FINE: Redefine Regression Task of DETRs as Fine-grained Distribution Refinement [ICLR 2025 Spotlight]
Yansong Peng, Hebei Li, Peixi Wu, Yueyi Zhang, Xiaoyan Sun, Feng Wu

DEIM: DETR with Improved Matching for Fast Convergence [CVPR 2025]
Shihua Huang, Zhichao Lu, Xiaodong Cun, Yongjun Yu, Xiao Zhou, Xi Shen

Real-Time Object Detection Meets DINOv3
Shihua Huang, Yongjie Hou, Longfei Liu, Xuanlong Yu, Xi Shen

RT-DETRv4: Painlessly Furthering Real-Time Object Detection with Vision Foundation Models
Zijun Liao^* , Yian Zhao^* , Xin Shan, Yu Yan, Chang Liu, Lei Lu, Xiangyang Ji, Jie Chen

(We will update this list as new contributions are published. If you have a work that belongs here, please open a pull request!)

The journey of real-time, end-to-end object detection is still in its exciting early stages. The success of the RT-DETR family proves that Transformer-based detectors are a dominant force in the making.

We look forward to seeing more innovations from the community. Let's work together to push the boundaries of real-time detection and build the next generation of intelligent systems!

Contributions are welcome! Feel free to submit issues or pull requests to help keep this repository up-to-date.