RF-DETR Object Detection API

Overview

RF-DETR is a real-time, transformer-based object detection model achieving SOTA performance on COCO benchmarks. This project exposes it as a production REST API via LitServe.

Key Features

• Transformer backbone — robust feature extraction without anchors or NMS
• End-to-end detection — single forward pass from image to bounding boxes
• High accuracy — SOTA on COCO and standard detection benchmarks
• LitServe serving — scalable, batching-ready inference endpoint

Get Started

cd rfdetr-object-detection
pip install -r requirements.txt
python server.py
curl -X POST http://localhost:8000/predict -F "[email protected]"