Skip to content

Latest commit

 

History

History
23 lines (15 loc) · 996 Bytes

README.md

File metadata and controls

23 lines (15 loc) · 996 Bytes
Raw

YOLOv4 conversion to OpenVINO (IR)

YOLOv4 -> ONNX

Script for conversion to ONNX format is available in tensorrt directory.

python yolo_to_onnx.py -c 80 -m ./yolov4 -o ./yolov4.onnx

ONNX -> OpenVINO (IR)

Conversion from ONNX to Intermediate Representation (IR) requires usage of OpenVINO toolkit with Model Optimizer package. It is a part of tool and is in <OPENVINO PATH>/deployment_tools/model_optimizer/ directory. Before running mo.py script it is necessary to source toolkit with command source <OPENVINO PATH>/bin/setupvars.sh .

# convert ONNX to OpenVINO IR with float32 weights
python mo.py --input_model ./yolov4.onnx --model_name yolov4_fp32 --data_type FP32 --batch 1

# convert ONNX to OpenVINO IR with float16 weights
python mo.py --input_model ./yolov4.onnx --model_name yolov4_fp16 --data_type FP16 --batch 1