How does Yolo2 create anchor boxes with only two points?

Question

Looking at YOLO scripts, it seems only two points are needed for anchor points.

ANCHORS = [0.57273, 0.677385, 1.87446, 2.06253, 3.33843, 5.47434, 7.88282, 3.52778, 9.77052, 9.16828]

If they represent height and width, what about the starting coordinates?

Answer 1

Anchor boxes are usually intended to be base-shapes of objects in a dataset, and exist without having a specific position. The way the anchor boxes are used depends on the network.

In YOLO there should be a grid on the feature space of the image. The anchor boxes will be placed on the intersection points of the grid to check whether an object of that shape exists in that position.