Table of Contents

• Darknet Object Detection Framework and YOLO
• Papers
• General Information
  • Darknet Version
• Pre-trained Weights
  • People-R-People Pre-trained Weights
  • MSCOCO Pre-trained Weights
• Building
  • Google Colab
  • WSL
  • Linux CMake Method
  • Windows CMake Method
  • Docker
• Using Darknet
  • CLI
  • Training
• Other Tools and Links
• Roadmap
  • Short-term goals
  • Mid-term goals
  • Long-term goals

Darknet Object Detection Framework and YOLO

Darknet is an open source neural network framework written in C, C++, and CUDA.

YOLO (You Only Look Once) is a state-of-the-art, real-time, object detection system, which runs in the Darknet framework.

• 2025-08: Darknet/YOLO repo has moved to Codeberg.org/CCodeRun/darknet/
  • All commits are automatically mirrored from Codeberg to the older Hank.ai GitHub repo.
• See the Darknet/YOLO web site
• Please read through the Darknet/YOLO FAQ
• Join the Darknet/YOLO discord server

Papers

• Paper YOLOv7
• Paper Scaled-YOLOv4
• Paper YOLOv4
• Paper YOLOv3

General Information

The Darknet/YOLO framework continues to be both faster and more accurate than other frameworks and versions of YOLO.

This framework is both completely free and open source. You can incorporate Darknet/YOLO into existing projects and products -- including commercial ones -- without a license or paying a fee.

Darknet V3 ("Jazz") released in October 2024 can accurately run the LEGO dataset videos at up to 1000 FPS when using a NVIDIA RTX 3090 GPU, meaning each video frame is read, resized, and processed by Darknet/YOLO in 1 millisecond or less. Additional performance improvements have been implemented in Darknet V4 and the upcoming Darknet V5, without compromising the quality of detections.

Please join the Darknet/YOLO Discord server if you need help or you want to discuss Darknet/YOLO and related tools: https://discord.gg/MQw32W9Cqr

The CPU version of Darknet/YOLO can run on simple devices such as Raspberry Pi, cloud & colab servers, desktops, laptops, and high-end training rigs. The GPU version of Darknet/YOLO requires either a CUDA-capable GPU from NVIDIA, or a ROCm-capable GPU from AMD.

Darknet/YOLO is known to work on Linux, Windows, and Mac. See the building instructions below.

Darknet Version

• The original Darknet tool written by Joseph Redmon in 2013-2017 did not have a version number. We consider this version 0.x.
• The next popular Darknet repo maintained by Alexey Bochkovskiy between 2017-2021 also did not have a version number. We consider this version 1.x.
• The Darknet repo sponsored by Hank.ai and maintained by Stéphane Charette starting in 2023 was the first one with a version command. From 2023 until late 2024, it returned version 2.x "OAK".
  • The goal was to try and break as little of the existing functionality while getting familiar with the codebase.
  • Re-wrote the build steps so we have 1 unified way to build using CMake on Windows, Linux, and Mac.
  • Converted the codebase to use the C++ compiler.
  • Enhanced chart.png while training.
  • Bug fixes and performance-related optimizations, mostly related to cutting down the time it takes to train a network.
  • The last branch of this codebase is version 2.1 in the v2 branch.
• The next phase of development started in mid-2024 and was released in October 2024. The version command now returns 3.x "JAZZ".
  • Removed many old and unmaintained commands.
    • You can always do a checkout of the previous v2 branch if you need to run one of these commands. Let us know so we can investigate adding back any missing commands.
  • Many performance optimizations, both when training and during inference.
  • Legacy C API was modified; applications that use the original Darknet API will need minor modifications: https://darknetcv.ai/api/api.html
  • New Darknet V3 C and C++ API to make it easier for developers to write applications that use Darknet/YOLO: https://darknetcv.ai/api/api.html
  • New apps and sample code in src-examples: https://darknetcv.ai/api/files.html
• The next release was in early 2025. The version command now returns 4.x "SLATE".
  • Added support for AMD GPUs using ROCm.
    • Still need to add support for MIOpen.
  • All printf() and std::cout calls have all been replaced so Darknet/YOLO logging messages can easily be redirected.
• The next release was in August 2025. The version command now returns 5.x "Moonlit".
  • Move source code repo from github to codeberg.
  • More performance optimizations.
  • Use of OpenBLAS when building CPU-only versions of Darknet.
  • Support for Profile-Guided Optimization.
  • ONNX export functionality. [experimental]
  • JAVA language bindings. [incomplete, in-progress]
• The most recent release is v5.1, which continues the "Moonlit" name introduced in v5.0.
  • Released in December 2025.
  • The ONNX export tool now includes the necessary nodes to export both "confs" and "boxes".
  • The mAP function was re-written.
  • Small number of new performance optimizations.

Pre-trained Weights

People are generally expected to train their own network. But pre-trained weights are also popular. These are weights which someone else has trained and made available for free on the internet. This is sometimes convenient when first installing Darknet/YOLO, since the software can be quickly tested without having to train a new neural network.

• People-R-People Pre-trained Weights (2 classes, person and head)
• MSCOCO Pre-trained Weights (80 classes covering a large range of objects, such as person, backpack, chair, clock, ...)
• There are several other simple datasets and pre-trained weights available for testing Darknet/YOLO, such as LEGO Gears and Rolodex. See <a target="_blank" href="https://www.ccoderun.ca/programming/yolo_faq/#datasets">the Darknet/YOLO FAQ</a> for details.

People-R-People Pre-trained Weights

The People-R-People pre-trained weights are used to find people. This dataset only has 2 classes: person and head.

You can download these pre-trained weights directly from C Code Run's web site.

MSCOCO Pre-trained Weights

Several popular versions of YOLO were pre-trained for convenience on the MSCOCO dataset. This dataset has 80 classes, which can be seen in the text file cfg/coco.names.

The MSCOCO pre-trained weights can be downloaded from several different locations, and are also available for download from this repo:

• YOLOv2, November 2016
  • YOLOv2-tiny
  • YOLOv2-full
• YOLOv3, May 2018
  • YOLOv3-tiny
  • YOLOv3-full
• YOLOv4, May 2020
  • YOLOv4-tiny
  • YOLOv4-full
• YOLOv7, August 2022
  • YOLOv7-tiny
  • YOLOv7-full

The MSCOCO pre-trained weights are provided for demo-purpose only. The corresponding .cfg and .names files for MSCOCO are in the cfg directory. Example commands:

wget --no-clobber https://github.com/hank-ai/darknet/releases/download/v2.0/yolov4-tiny.weights
darknet_02_display_annotated_images coco.names yolov4-tiny.cfg yolov4-tiny.weights image1.jpg
darknet_03_display_videos coco.names yolov4-tiny.cfg yolov4-tiny.weights video1.avi
DarkHelp coco.names yolov4-tiny.cfg yolov4-tiny.weights image1.jpg
DarkHelp coco.names yolov4-tiny.cfg yolov4-tiny.weights video1.avi

Building

The various build methods available in the past (pre-2023) have been merged together into a single unified solution. Darknet requires C++17 or newer, OpenCV, and uses CMake to generate the necessary project files.

[!NOTE] You do not need to know CMake, C++, C, Python, or any other programmaing language to build, install, nor run Darknet/YOLO. Some advanced topics -- such as using the Darknet/YOLO API -- does require programming experience.

Select one of the following build types:

• Google Colab
• WSL
• Linux
• Windows
• Docker

[!CAUTION] Beware if you are following old tutorials with more complicated build steps, or buil