hank-ai/darknet

Darknet/YOLO object detection framework

What it solves

Darknet is an open-source neural network framework designed for high-performance object detection. It primarily serves as the engine for YOLO (You Only Look Once), a real-time system that identifies and locates objects within images and videos with high speed and accuracy.

How it works

Written in C, C++, and CUDA, Darknet is built for efficiency. It supports both CPU and GPU acceleration (NVIDIA via CUDA and AMD via ROCm). The framework allows users to either use pre-trained weights for common objects (like the MSCOCO dataset) or train their own custom neural networks. Recent versions have introduced performance optimizations, including Profile-Guided Optimization and OpenBLAS for CPU-only builds, and experimental support for ONNX export.

Who it’s for

  • Developers looking to integrate real-time object detection into commercial or open-source products via C/C++ APIs.
  • Researchers and AI practitioners who need a fast, lightweight framework for training and deploying vision models.
  • Users who want to run object detection on a variety of hardware, from Raspberry Pi to high-end NVIDIA/AMD GPUs.

Highlights

  • Extreme Speed: Capable of processing video frames in 1ms or less (up to 1000 FPS) on high-end GPUs.
  • Broad Hardware Support: Runs on Linux, Windows, and Mac, supporting NVIDIA, AMD, and CPU-only configurations.
  • Permissive Licensing: Completely free and open source for use in commercial projects without fees.
  • Flexible Deployment: Provides a unified CMake build system and installation packages (.deb, .rpm, .exe) for easier setup.

Related

  • Project
    ultralytics/ultralytics

    A comprehensive computer vision framework featuring YOLO models for real-time object detection, segmentation, classification, and pose estimation.

  • Project
    yzfzzz/depth-detect

    A high-performance C++ and TensorRT framework that fuses YOLO object detection with monocular/binocular depth estimation to track objects and estimate their motion states.

  • Project
    Geekgineer/YOLOs-CPP

    A production-ready C++ inference engine that provides a unified API for the entire YOLO family, supporting tasks from object detection to metric depth estimation.

  • Project
    aparsoft/yolo-streamlit-detection-tracking

    A Streamlit-based vision studio for real-time object detection, segmentation, pose estimation, and tracking using YOLO26 and YOLO World v2.

  • Project
    RizwanMunawar/yolov7-object-tracking

    A computer vision project that integrates YOLOv7 and YOLOv8 for real-time object detection and tracking across video streams from various sources.