Mastering Computer Vision with TensorFlow 2.x

Bounding box regression

Bounding box regression predicts the location of an object within an image. After the SVM, a linear regression model is developed to predict the location and size of the bounding box detection window. The bounding box of an object is defined by four anchor values, [x, y, w, h], where x is the x coordinate of the bounding box origin, y is the y coordinate of the bounding box origin, w is the width of the bounding box, and h is the height of the bounding box.

The regression technique attempts to minimize the errors in bounding box prediction by comparing the predicted value with the ground truth (target) values by adjusting each of the four anchor values.