I really like the speed and accuracy of your human detection model, the skeletons and 3D bounding boxes seem very useful. We need similar functionality in our project, but for animals instead of people. Do you have some sort of manual for creating a custom model like this? Because I failed to find one, and if not, could I ask you to create one?
Hello,
you can train a (some) model(s) with CustomVision on Microsoft Azure servers and then you can export an *.onnx file.
This export give you a *.cs file and a *.py file. Use the *.cs file into your project and replace the basic image input by your video frame.
Wouldn’t that be just a 2D detection model though? ZED 2 models are stereoscopic and that is what I want to achieve. @PrivHate
[ONNX] Accurate my own detection · Issue #15 · stereolabs/zed-csharp-api (github.com)
People are a primary focus for the object detection in the ZED SDK yes, the precision, in that case, partly comes from the dataset. There are more humans than other classes of objects and we focus our work on this accuracy.