Summary

Object recognition and detection is capable of identifying several objects within an image, to draw bounding boxes around those objects and predict the types of object they are.

The process of labeling the bounding boxes and their labels has been explained, but not in depth, due to the huge process required. Instead, we used state-of-the-art models to recognize and detect those objects.

YOLOV3 was the main model used in this chapter. OpenCV was used to explain how to run an object detection pipeline using its DNN module. ImageAI, an alternative library for object detection and recognition, has shown its potential for writing an object detection pipeline with a few lines and easy object customization.

Finally, the ImageAI object detection pipeline was put into practice by using a video, where every frame obtained from the video was passed through that pipeline to detect and identify objects from those frames and show them using Matplotlib.