Original Bounding Box Coordinates to Resized Image Coordinates

I am currently trying to import new predictions using the python SDK.

Currently I have the bounding box coordinates in COCO format for a rectangle label based on the original image dimensions.

When trying to import predictions with the original bbox coordinates, they are now too big as the image has been resized. I see that annotations for Images are saved as percentages of the original size of the image from 0-100.

I am able to convert the x_min and y_min values but am having difficulty calculating the new values for width and height.