why add strides_expanded / 2?

[pengcheng001]

📚 Documentation

In the CenterNet2/centernet/modeling/dense_heads/centernet.py file, when calculating the distribution of center points, centers_discret = ((centers_expanded / strides_expanded).int() * strides_expanded).float() + trides_expanded / 2 # M x N x 2, why add strides_expanded / 2?

[xingyizhou]

Thank you for the question and sorry for the delayed response. This follows RetinaNet to put the point in the center of each grid. This is useful when the output stride is large (i.e., in FPN the maximal stride is 128).