dg_export_int8_output.py

from ultralytics import YOLO

import argparse

def parser_arguments():
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', type=str, default='', help='initial weights path')
    parser.add_argument('--format', type=str, default='tflite', help='export format')
    parser.add_argument('--quantize', action='store_true', help='int8 export')
    parser.add_argument('--data', type=str, default='coco128.yaml', help='dataset.yaml path')
    parser.add_argument('--imgsz', '--img', '--img-size', type=int, default=640, help='train, val image size (pixels)')

    return parser.parse_args()

if __name__ == '__main__':

    args = parser_arguments()
    # Create a dictionary of kwargs
    kwargs = vars(args)
    print(kwargs)

    model = YOLO(args.weights)

    # success = model.export(format=args.format, imgsz=args.imgsz, data=args.data, int8=args.quantize, separate_outputs=True, export_hw_optimized=True, uint8_io_dtype=True, max_ncalib_imgs=100)
    
    #uint8_io_dtype=True
    # success = model.export(format=args.format, imgsz=args.imgsz, data=args.data, int8=True, separate_outputs=True, export_hw_optimized=True, uint8_io_dtype=True, max_ncalib_imgs=100)
    
    #uint8_io_dtype=False
    success = model.export(format=args.format, imgsz=args.imgsz, data=args.data, int8=True, separate_outputs=True, export_hw_optimized=True, uint8_io_dtype=False, max_ncalib_imgs=100)