Why predict the coordinates of bounding boxes in an Image?

[AbdullahMastan01]

Timestamp of Video (Part 1): 3:54:54

You Mentioned that the regression problem could be used to predict the coordinates of a bounding box in an image.... kindly elaborate

From what I know, the CV model uses trained weights to identify whether the aggregation of pixels in an image form a specific object. Once it the prediction is True, the model just draws the bounding box around the object along with some padding.

Then why predict the bounding box using regression?

Thank You.