Query about the input image size

[NTUZZH]

Hi there,

I would like to ask how to define the input image size for the model?

Let's say If my original image is 1920 * 1080, will I get a better segmentation result with input image size of 1920 * 1920 compared with 1024 * 1024? (refer to this line: parser.add_argument("--imgsz", type=int, default=1024, help="image size"))

Thanks guys!

-ZZH