Hi, I’m currently trying to inference a yolov8s_seg model that I trained, but the thing is when converting to hef file, there is no post nms and I can’t find anything related to the configuration for any segmentation models.
So is there a documentation on how to do this post segmentation or is the a config file for hailomz conversion
Hi @Alterra_Onix
We developed a python SDK to simplify working with Hailo devices and integrated segmentation postprocessor. You can check these links for more information:
But it seems like this only works for pretrained model in hailo model zoo right?
What I want is inference using my own trained model
@Alterra_Onix
You can port your own pre-compiled model. Every model has a json file associated with it in which you need to specify the number of classes (from your output it seems you have 12 classes) and a file with labels. You can see details of model json in our user guide: User Guide 1 Hailo World: Running Your First Inference on a Hailo Device Using DeGirum PySDK. Please let me know if encounter any issues in getting your model to run.
Hi @Alterra_Onix
We published a guide to port your custom segmentation model to PySDK: User Guide 4: Simplifying Instance Segmentation on a Hailo Device Using DeGirum PySDK