How to create post processing steps for the yolo-obb model for Hailo

[saurabh-git-dev]

Hi, I am working with Hailo8l on Raspberry Pi 5. The Hailo compiler does not compile all the operations from the Yolo onnx models. We have to define end nodes; the last few operations are not compiled. So we have to manually build those processing steps and run on the cpu.
But this is something quite complicated for me as a beginner in ML. And I am not sure, how to build the post-processing part for yolo-obb models. The Hailo team provides a few predefined post-processing functions. But not for all models.
So, I am finding a way to build post-processing steps.

YOLO v8n OBB


There are a total of 9 output layers(Marked in red color) that i need to process.

Hailo returns the conv raw outputs. And we have to post-process the result. And I want to write a method to process rest operations. But due to a lack of information, I am unable to start.

So, I am looking for more information and resources. I have also asked in the Hailo Community but no solution yet.
I am also willing to pay if someone is interested in building the post-processing part for me.

[UltralyticsAssistant]

👋 Hello @saurabh-git-dev, thank you for reaching out and sharing your project details! 🚀

Navigating post-processing for models like YOLO-OBB can indeed be challenging, especially when dealing with hardware-specific constraints like the Hailo compiler. While waiting for assistance, I recommend checking our Docs for insights on YOLO models, which might offer some foundational understanding.

If this is related to code 🐛, please provide a minimum reproducible example to help us better understand the context and assist you efficiently. This example can include any scripts or configurations specific to your current setup.

For community support, you've taken great steps by also reaching out in the Hailo Community. You can further engage with experts and enthusiasts in our Discord 🎧 where you might discover more tailored advice.

Additionally, to ensure you're up to date with the latest features and fixes, please upgrade to the latest ultralytics package:

pip install -U ultralytics

Here's more information on running environments that might help streamline your development:

• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

This is an automated response, and an Ultralytics engineer will assist you soon. In the meantime, feel free to engage with the resources mentioned above. 😊

Status

If this badge is green, all Ultralytics CI tests are operational, ensuring that all modes and tasks are functioning correctly on different platforms.

[glenn-jocher]

@saurabh-git-dev thank you for reaching out. For guidance on post-processing YOLO-OBB models, I recommend checking the Ultralytics documentation and community forums for insights. Unfortunately, we can't offer private support or custom development services.