How does an AI turn a picture into boxes labelled “person” or “dog”? DxL’s YOLO.EXE makes the intermediate steps visible: divide the image into a grid, predict boxes and classes, then remove overlapping guesses.

The explanation starts with the original YOLO architecture. A separate browser demo lets you supply an image and adjust detection settings with newer models. Seeing which boxes disappear when you raise confidence makes the tradeoff much easier to grasp.

An inviting introduction to object detection, with an important distinction: newer YOLO versions do not all work exactly like the historical example.