加载中
正在获取最新内容，请稍候...
RT-DETR: Real-Time Detection Transformer | DETRs Beat YOLOs on Real-time Object Detection
Official implementation of RT-DETR, a Real-Time DEtection TRansformer model that aims to outperform traditional YOLO-based methods in real-time object detection tasks. Supports both PaddlePaddle and PyTorch frameworks.
Python
Added on 2025年6月21日3,811
Stars446
ForksPython
LanguageProject Introduction
Summary
RT-DETR is a cutting-edge object detection model designed for real-time performance. It adapts the DETR architecture for speed and efficiency, challenging the dominance of YOLO models in the real-time detection space with competitive accuracy and faster inference times.
Problem Solved
Addresses the need for high-accuracy, real-time object detection models that can leverage the benefits of Transformer architectures while remaining computationally efficient for deployment in various real-world applications.
Core Features
Real-Time Performance
Achieves state-of-the-art real-time inference speeds on various hardware platforms.
Efficient DETR Architecture
Based on the efficient Detection Transformer architecture for end-to-end object detection.
Cross-Framework Support (PaddlePaddle & PyTorch)
Provides official implementations and pre-trained models for both PaddlePaddle and PyTorch.
High Accuracy
Demonstrates competitive or superior accuracy compared to leading YOLO models.
Tech Stack
PaddlePaddle
PyTorch
DETR
Transformer
Object Detection
Python
Usage Scenarios
Due to its real-time performance and high accuracy, RT-DETR is suitable for a wide range of applications where rapid and reliable object detection is crucial:
Edge Computing & Mobile Applications
Details
Deploying models on edge devices or embedded systems for immediate analysis of visual input, such as in drones, mobile devices, or IoT cameras.
User Value
Enables powerful AI capabilities directly on resource-constrained hardware without relying on cloud processing.
Robotics and Autonomous Systems
Details
Identifying and tracking objects in dynamic environments for navigation, control, or interaction, such as in autonomous vehicles or robotic arms.
User Value
Provides the crucial real-time environmental perception needed for safe and effective operation.
Real-Time Monitoring and Surveillance
Details
Analyzing live video streams from surveillance cameras or industrial monitoring systems for security, safety, or quality control purposes.
User Value
Allows for instant anomaly detection, event triggering, and automated reporting.
Recommended Projects
You might be interested in these projects
quarkusioquarkus
Quarkus is a Kubernetes-native Java framework tailored for GraalVM and HotSpot, crafted from best-of-breed Java libraries and standards. It's designed to enable developers to create high-performance, lightweight applications quickly.
Java
146492889
mpv-playermpv
MPV是一个免费、开源、跨平台的媒体播放器，以其极简的界面、强大的命令行控制、广泛的格式支持和灵活的脚本能力而闻名。它是MPlayer和mplayer2的一个分支，专注于提供高品质的视频输出和可定制的用户体验。
C
311283053
MetaCubeXmihomo
A Python library using Pydantic for parsing Honkai: Star Rail game data from the Mihomo API, providing structured access to player profiles, characters, relics, and more.
Python
211303090