Artificial Intelligence Programming with Python

CHAPTER 7Object Detections and Image Segmentations

“AI doesn't have to be evil to destroy humanity—if AI has a goal and humanity just happens to come in the way, it will destroy humanity as a matter of course without even thinking about it, no hard feelings.”

—Elon Musk (American entrepreneur, industrial designer, and engineer)

CHAPTER OUTLINE

• 7.1 Introduction
• 7.2 Object Detections with Pretrained Models
• 7.3 Object Detections with Custom Trained Models
• 7.4 Object Tracking
• 7.5 Image Segmentation
• 7.6 Background Removal
• 7.7 Depth Estimation
• 7.8 Augmented Reality
• 7.9 Summary
• 7.10 Chapter Review Questions

7.1 Introduction

Object detection is another important aspect of image analysis or image processing. Different from image classification, which identifies the content of the whole image, object detection detects and locates different objects within an image, for example, to detect dogs and cats in an image or to detect cyclists, vehicles, and pedestrians in an image.

Traditionally, object detection is achieved by detecting colors, shapes, and contours, and you can typically implement this by using the popular OpenCV library. In Hollywood, people have long been using green-colored backgrounds during filming, which can then be easily replaced with different scenes later. However, this approach is suitable only for detecting simple objects.

A much more modern approach is to use deep learning neural network models, which are suitable for detecting more complex objects. ...