Download
RF-DETR is an open-source computer vision framework that implements a real-time object detection and instance segmentation model based on transformer architectures. Developed by Roboflow, the project builds upon modern vision transformer backbones such as DINOv2 to achieve strong accuracy while maintaining efficient inference speeds suitable for real-time applications. The model is designed to detect objects and segment them within images or video streams using a unified detection pipeline. RF-DETR emphasizes strong performance across both accuracy and latency benchmarks, allowing developers to deploy high-quality detection models in applications that require immediate processing such as robotics, autonomous systems, and industrial inspection. The repository includes Python packages, training scripts, and model configurations that enable researchers and engineers to train and deploy detection models on custom datasets.
Features
- Transformer-based architecture for object detection
- Support for instance segmentation and detection tasks
- Built on DINOv2 vision transformer backbone
- Real-time inference optimized for performance and latency
- Python package and training pipeline for custom datasets
- Benchmarks demonstrating strong accuracy on standard datasets
Project Samples
