I am using the images directly to show video and detect an object in it. Problem is that either I\'ll have to hardcode the ROI or get a bounding box around it using
ROI