Using TensorRt to perform inference accuracy loss is too much. When generating the engine, the data type uses DataType::kFLOAT. It is not clear whether it is the format problem of opencv to read the image. I hope you can help me.

Hi @429087397,
Are you using caffe parser here?
If yes, we recommend you to use ONNX parser.
Also please share your model and script along with the below set of details so that we can assist you better.

o Linux distro and version
o GPU type
o Nvidia driver version
o CUDA version
o CUDNN version
o Python version [if using python]
o Tensorflow and PyTorch version
o TensorRT version