Before learning about non-maximal suppression, we’ll need to learn about intersection over union or IOU, which is a technique used in non-maximal suppression to compare how good two different bounding boxes are for a given object. It’s easiest to see how to calculate IOU in an example. Take these two bounding boxes. We define the intersection over union for the two boxes to be the ratio of the area of intersection to the area of Union. The area of intersection between the two boxes is marked by the green rectangle and it’s just the area where these boxes overlap. The area of the union is denoted by the purple area and is the total area of the boxes if they were smooched into one bigger unified shape. The area of intersection is 900 square pixels, while the union of the boxes is 3000 square pixels. This means that the IOU of the boxes is 900 over 3000 or 0.3. Now, imagine you’re comparing a ground-truth box to predicted box, what IOU value would indicate a good match?