When i use yolov5 scripts to predict yolov8 model i am getting abnormal scores, predictions

[ajithvcoder]

I trained a custom model and i am using your yolov5 scripts . The output tensor when i use yolov8 model is not matching with python version should i need to update your yolov5 script for yolov8 ? Also the scores, labels, boxes are going to 10000, 2000 ranges

[craft-zhang]

Of course, the project is pretty ugly right now. The support for different versions of yolo is also messy. But thank you very much for your contribution and support.

[ajithkumarmcw]

thanks for your reply could you give me any links which would work for float32 model for yolov8

[craft-zhang]

currently the scores, labels, and boxes are hard coded in the YOLO::run() method and for the input data format you can ref

tensorflow-lite-cpp-examples/yolov5.cc

Line 217 in a4b9c02

} else if (_in_type == kTfLiteUInt8) {

[ajithvcoder]

okay thanks