How to convert YOLOv4 Pytorch-ONNX --> TensorRT engine INT8 mode with INT8 calibration and Dynamic Shape

NVIDIA-AI-IOT / yolo_deepstream

yolo model qat and deploy with deepstream&tensorrt

Apache License 2.0

534 stars 135 forks source link

How to convert YOLOv4 Pytorch-ONNX --> TensorRT engine INT8 mode with INT8 calibration and Dynamic Shape #11

Open vilmara opened 3 years ago

vilmara commented 3 years ago

Hi all, is there a sample that shows how to optimize a YOLOv4 Pytorch-ONNX to TensorRT engine INT8 mode with full INT8 calibration and dynamic input shapes?. I have generated the TensorRT engine INT8 mode in runtime with DeepStream but it doesn't generate the engine with dynamic input shapes. Also using trtexec it built the engine with static input shapes (default BS=1) and doesn't provide calibration capability.