At Hank.ai, we’re passionate about empowering organizations with cutting-edge AI technology, especially in the realm of revenue management. But our mission extends beyond proprietary solutions—we are also committed to supporting and advancing open-source initiatives that push the boundaries of innovation. Today, we’re thrilled to announce the release of Darknet V3 (codenamed “JAZZ”), a significant update to the widely-used Darknet/YOLO open-source object detection framework. Led by Stéphane Charette, this latest version brings exceptional performance enhancements, enhanced developer tools, and full backward compatibility with existing YOLO models.

Darknet/YOLO is an essential component of many computer vision projects, and Hank.ai is proud to sponsor this open-source community effort. By enhancing this project and contributing to its growth, we make it easier for developers, researchers, and companies to implement high-performance object detection technology into their applications.

“Darknet V3 demonstrates our commitment to advancing open-source computer vision technology,” said Stéphane Charette, lead developer of the project. “The performance improvements and new developer tools will make it easier than ever for organizations to implement robust object detection in their applications.”

Key Improvements in Darknet V3 (“JAZZ”)

  • Dramatic Performance Gains: Darknet V3 brings significant speed boosts—projects that previously ran at 280 FPS can now achieve an impressive 382 FPS. This represents a massive leap forward for organizations that require real-time processing capabilities.
  • Accelerated Image Processing: In this new version, loading 5000 images now takes just 38.1 seconds, compared to 48.2 seconds in the previous version. This time reduction can significantly impact workflows where image batch processing speed is crucial.
  • New Developer API: A comprehensive and well-documented API provides developers with seamless integration options for their applications. This improvement greatly simplifies working with the Darknet framework for both experienced users and newcomers.
  • Improved CLI: The command-line interface has been reimagined for better usability. It’s more intuitive, allowing for quicker commands, and is also accessible via the new API, bridging the gap between ease-of-use and advanced development.
  • Complete Backwards Compatibility: Darknet V3 ensures a smooth transition for existing users by maintaining full support for YOLO weights and configuration files. This means current users of the platform can upgrade without losing compatibility with their existing models.

A Commitment to Open Source

The Darknet framework remains completely open-source and free to use, with no special licenses or permits required for integration into your projects. By making this technology widely accessible, we aim to empower innovation at all levels.

Developers interested in exploring the new capabilities of Darknet V3 can find extensive documentation, example applications, and community support through the links below:

Hank.ai’s sponsorship of Darknet/YOLO underscores our belief that the future of technology lies in collaboration and open innovation. We’re excited to see what the developer community will build with Darknet V3, and we remain committed to supporting the growth of this incredible project.