MCPcopy Create free account
hub / github.com/CVCUDA/CV-CUDA / main

Function main

samples/operators/reformat.py:31–54  ·  view source on GitHub ↗

Reformat an image with CVCUDA.

()

Source from the content-addressed store, hash-verified

29
30
31def main() -> None:
32 """Reformat an image with CVCUDA."""
33 # docs_tag: begin_main
34 args: argparse.Namespace = parse_image_args()
35 # docs_tag: begin_read_image
36 input_image: cvcuda.Tensor = read_image(args.input)
37 height, width, channels = input_image.shape
38 # docs_tag: end_read_image
39
40 # docs_tag: begin_reformat
41 # 1. Perform a reformat operation on the image to CHW
42 chw_image: cvcuda.Tensor = cvcuda.reformat(input_image, "CHW")
43 assert chw_image.shape == (channels, height, width)
44
45 # 2. Perform a reformat operation on the CHW image back to HWC
46 hwc_image: cvcuda.Tensor = cvcuda.reformat(chw_image, "HWC")
47 assert hwc_image.shape == (height, width, channels)
48
49 # 3. Add a batch dimension to the image and reformat to NCHW
50 nhwc_image: cvcuda.Tensor = cvcuda.stack([input_image])
51 nchw_image: cvcuda.Tensor = cvcuda.reformat(nhwc_image, "NCHW")
52 assert nchw_image.shape == (1, channels, height, width)
53 # docs_tag: end_reformat
54 # docs_tag: end_main
55
56
57if __name__ == "__main__":

Callers 1

reformat.pyFile · 0.70

Calls 2

parse_image_argsFunction · 0.90
read_imageFunction · 0.90

Tested by

no test coverage detected