If using the tensorRT engine, the input should be NCHW.
Could you comment the line for the image type?
I am confuse at the process of NCHW and NHWC.
Thanks
OpenCV reads images into memory in NHWC format, while TensorRT expects images in NCHW format. The conversion is done in the code here as part of the pre-processing before running inference on the inputs.
If using the tensorRT engine, the input should be NCHW. Could you comment the line for the image type? I am confuse at the process of NCHW and NHWC. Thanks