search

Megvii-BaseDetection / YOLOX

YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with MegEngine, ONNX, TensorRT, ncnn, and OpenVINO supported. Documentation: https://yolox.readthedocs.io/
Apache License 2.0
9.47k stars 2.21k forks source link

Openvino C++ inference issue #531

Closed Dereck-Ting closed 3 years ago

Dereck-Ting commented 3 years ago

result I trained my own data with 6 classes and convert it to openvino xml file. I tested this xml and bin file under openvino python script and works good , but I got the weird results when tested in Openvino C++. Do I need to change something in c++ file besides class number?

GOATmessi8 commented 3 years ago

check the img input size in cpp file to be consistent with your converted model, and the num_class should also be changed

Dereck-Ting commented 3 years ago

check the img input size in cpp file to be consistent with your converted model, and the num_class should also be changeI

The input size in cpp file is consistent with the model, I used yolox_tiny and set the size to be both 416, but the result in cpp is still werid.