YOLO v11 Open Source

Clean, modular, Apache 2.0 licensed YOLOv11 implementation in pure PyTorch with multi-task support

🚀 YOLOv11 Reimagined

A complete, license-free reimplementation of YOLOv11 in pure PyTorch. Built for researchers and engineers who need transparent, hackable code without vendor lock-in.

🐍 Pure PyTorch 📜 Apache 2.0 ðŸŽŊ Multi-Task ⚡ Production Ready
🚧
December 2025 Update: Open-source weights are coming soon! Currently training all models from scratch on COCO for fully Apache 2.0 licensed weights.
ðŸŽŊ

Object Detection

80 COCO classes with multi-scale FPN for accurate bounding box predictions

🎭

Instance Segmentation

Pixel-perfect instance masks using prototype-based segmentation

ðŸĶī

Pose Estimation

17 keypoints with skeleton connections for human pose analysis

âœĻ Multi-Task Capabilities

ðŸŽŊ Detection
🎭 Segmentation
ðŸĶī Pose

🆚 Why Choose This Implementation?

Feature Ultralytics This Project
License AGPL-3.0 Apache 2.0 ✓
Commercial Use Requires paid license Free ✓
Dependencies Ultralytics package Pure PyTorch ✓
Architecture YOLOv11 100% Compatible ✓
Code Transparency Abstracted Fully Readable ✓

🌟 Ready to Get Started?

Clone the repository and start building with license-free YOLOv11 today!

View on GitHub

📚 References

  1. YOLOv11-pt by @jahongir7174 — Initial reference implementation
  2. Ultralytics — Original YOLO architecture

Built for the Open Source Community 🌍