RT-DETR conversion to int8

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

I am working on convertion RT-DETR to int8 precision. I succesfully obained .onnx file, and using trtexec i am able to convert to FP32 and FP16. But for int8, I need to obtain calibration.

I already check tensorrt nvidia official documentation, but this did not help.

I am using Jetson 5.1.1, TensorRt 8.5.2.2

2

Found a repo that might be helpful, I’m trying out. But ran in some issues also when converting to int8. Here’s the repo: DeepStream-Yolo/docs/INT8Calibration.md at master · marcoslucianops/DeepStream-Yolo · GitHub. Please let me know if you can figure it out T.T