From the Stereo neural inference
the example gen2-triangulation uses a facial detector and a facial landmark detector.
I have a YOLOv8 trained custom object detector, how can I create/provide the custom object landmark detector?
Stereo neural inference - Triangulation - Landmarks
- Edited
Hi kle
You could use it the same way. The problem is that you need to train your own (or use pretrained) neural network for landmark regression. Eg. the example you are referencing uses https://docs.openvino.ai/2023.0/omz_models_model_landmarks_regression_retail_0009.html model which outputs the landmarks. These are specific to faces; other objects would need custom training.
I'd suggest you look up some tutorials or papers on how to such a model is trained. Once that is done, the integration should be relatively simple.
Thanks,
Jaka