Mastering Computer Vision with TensorFlow 2.x

Selective search

The primary challenge in object detection is to find the precise location of the object in an image. Multiple objects in an image at different spatial orientations make it hard to find the boundary of the object in an image. For example, an object can be covered and only partially visible—an example is a person standing behind a car; we can see the car and the person's body above the car. A selective search is used to solve this problem. It divides the whole image into many segmentation regions. It then uses a bottom-up approach to combine similar regions into larger regions. The selective search uses the generated region to find the object's location. The selective search uses a greedy algorithm to iteratively group regions ...