Can YOLOv8 Simultaneously Support Rotated Object Detection (OBB) and Keypoint Detection? about ultralytics HOT 4 OPEN

Hello! Great to see your interest in extending YOLOv8 capabilities for your project. Currently, YOLOv8 does not natively support a single model that handles both Oriented Bounding Boxes (OBB) and keypoint detection simultaneously. Each task is typically handled by distinct model architectures due to their different output requirements.

However, you can approach this challenge by:

1. Sequential Processing: First use an OBB model to detect objects and then run a keypoint detection model on the regions defined by the OBBs.
2. Custom Model Development: Modify the YOLOv8 architecture to output both keypoints and OBBs. This would involve significant changes to the model's head to predict additional keypoints along with the angle for rotation.

For the second approach, you might need to dive deep into the model's architecture and training pipeline, adjusting the loss functions to handle both tasks effectively.

No existing implementations directly combine these two in YOLOv8, but exploring related research papers or similar tasks might give you some insights on integration.

If you decide to modify the YOLOv8 codebase, consider the impact on training complexity and inference performance. Good luck with your project, and feel free to reach out if you have more questions! 🚀

from ultralytics.

Thank you for your helpful advice. I’m now following your suggestion to use OBB detection followed by keypoint detection. However, I still have a query I hope you can clarify for me. For identifying a simple triangle, should I label the three keypoints with the same tag, or should each keypoint have a unique label? I don't have any sequence or position constraints; I just need the keypoints that form the triangle. The order does not matter to me, as long as the model can predict these three points. Once again, thank you for your assistance.

from ultralytics.

You're welcome! For your triangle keypoints, you can label all three keypoints with the same tag since the order doesn't matter to you. This way, the model will learn to detect the three points that form the triangle without worrying about their sequence. Happy coding! 😊

from ultralytics.

👋 Hello there! We wanted to give you a friendly reminder that this issue has not had any recent activity and may be closed soon, but don't worry - you can always reopen it if needed. If you still have any questions or concerns, please feel free to let us know how we can help.

For additional resources and information, please see the links below:

• Docs: https://docs.ultralytics.com
• HUB: https://hub.ultralytics.com
• Community: https://community.ultralytics.com

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐

from ultralytics.