Loading pretrained detection models on the OpenMV?

I’m working on a project to detect cars using the OpenMV Cam.

I have done similar detections previously using Edge Impulse, which required collecting training images, labeling them, and then training the model on Edge Impulse. The trained model is then exported and loaded onto the OpenMV.

Since my project is detecting cars, and there are already models that exist to detect cars, is there a way to find these models and use them, so I can skip the data collection/labeling/training that would otherwise be required?

For example, the YOLO5 model is already trained to detect cars. Is there some way I can just use this model as is without training it further?

This is a question better suited to be asked on the Edge Impulse forms. You’d have to use their FOMO model which runs on our device. I’m not sure if YOLO can be transferred to a FOMO model. But if it can be, then the answer would be yes.

1 Like