Use my own object detection model.

OpenMV Products
1

I have trained a YOLO model using the Ultralytics framework, quantized it, and converted it to TFLite format. However, when I use it with the tf.detect function in OpenMV, it shows an error: “OSError: not a detect model or does not have post-processing node.” I suspect the issue is with the model’s output format. How should I modify it to meet the requirements?

2

Hi, tf.detect() was removed from our firmware a while back. We do not support that feature anymore.

Please see this: OpenMV Firmware v4.5.6 and up TensorFlow Porting Guide - OpenMV Products - OpenMV Forums