Mapping of cells output with actual inference image

[akshay-bothra]

Hi,
I am using inference for extracting cells objects. The output of detect is cropped and passed as an input to recognize. Now, recognize produces the cell objects for cropped table image. I want to map these cell bboxes on the actual image that was fed into the detection model. If you could provide some guidance on How should I map these to actual image input?