YOLOv8 Pose Custom Model – Inference Check and Optimization Tips

General
, ,
1

Hi,

I have generated a custom YOLO Pose model with a different number of keypoints compared to the standard YOLOv8 Pose, and it is trained on a single class (person).
What is the fastest way to visualize the inference results to check if the model conversion worked correctly?

Also, do you have any suggestions on how to preserve performance and avoid errors in case the model does not perform well? I’m looking for strategies to fine-tune the model while retaining the same input/output structure and avoiding any loss in accuracy.
For optimization phase, I used 170 images and trained for 10 epochs.

Thanks in advance for your support.

2

Have a look at our Application Example repository.

GitHub - Hailo-Application-Code-Examples - Python - Pose estimation

GitHub - Hailo-Application-Code-Examples - CPP - Pose estimation

I recommend to work trough the model optimization tutorials in the Hailo AI Software Suite Docker. Run the following command to start a Jupyter Notebook server with notebook for each step of the model conversion workflow.

hailo tutorial
1 Like