R

Rtdetr R34vd

Developed by PekingU
RT-DETR is the first real-time end-to-end object detection Transformer model, achieving high-speed and high-precision detection through an efficient hybrid encoder and query selection mechanism
Downloads 512
Release Time : 2/24/2024

Model Overview

Real-Time Detection Transformer (RT-DETR) is an end-to-end object detector that eliminates Non-Maximum Suppression (NMS), significantly improving speed while maintaining accuracy

Model Features

Efficient Hybrid Encoder
Rapidly processes multi-scale features by decoupling intra-scale interaction and cross-scale fusion
Uncertainty Minimization Query Selection
Provides high-quality initial queries for the decoder to improve detection accuracy
Flexible Speed Adjustment
Supports speed adjustment by tuning the number of decoder layers without retraining
NMS-Free Design
End-to-end architecture eliminates the Non-Maximum Suppression bottleneck in traditional object detection

Model Capabilities

Real-time object detection
Multi-scale object recognition
End-to-end prediction
High-precision bounding box regression

Use Cases

Intelligent surveillance
Public space crowd analysis
Real-time detection of crowds and objects in public spaces like airports and stations
Efficient processing capability at 108FPS@T4 GPU
Sports analysis
Live match analysis
Detecting player and ball positions in football matches
High-precision tracking of moving objects
Autonomous driving
Road object detection
Real-time identification of vehicles, pedestrians, and other targets on the road
54.3% AP on COCO
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase