Currently, training a YOLOv8 model produces .pt model files which cannot be directly used with OpenCV's cv2.dnn functions. We need a script that can convert these existing models to .onnx format so that we can then run inference on them. The functionality of this script could eventually be structured to work within cv2dnn.py.
Currently, training a YOLOv8 model produces
.pt
model files which cannot be directly used with OpenCV'scv2.dnn
functions. We need a script that can convert these existing models to.onnx
format so that we can then run inference on them. The functionality of this script could eventually be structured to work withincv2dnn.py
.