Object Detection and Tracking: Exploring a Deep Learning Approach and Other Techniques

Abstract

Object detection and tracking have a wide range of uses, for example, in security and surveillance systems to deter and investigate crimes, for traffic monitoring, and for communication through video sharing. In a smart city, data is continuously being obtained through various means. In terms of video data, the data collected through cameras and other digital devices need to be analyzed to derive useful information from it. Hence, the concept of object detection and tracking comes into play. This chapter looks into developing various frameworks for object detection and tracking in the context of video data. We will be working with a database of 1,939 pencil images. These images will be used to train a neural network that performs image classification tasks. Various object detection methods are implemented such as Convolution Neural Network (CNN) image classification, Canny edge detection, object detection using the Haar Cascade classifier, and background subtraction. The experiments are carried out with Python programming language, TensorFlow, and OpenCV library.<br>

Design and development of a quadruped home surveillance robot

“…The Raspberry Pi's camera serial interface (CSI) port accepts a direct plug from the Raspberry Pi Camera Board. It can produce images with a clear 5-megapixel (MP) quality or record 1080p high dynamic (HD) video at 30 frames per second [23]. The board itself is ideal for portable or various uses where weight and dimension are crucial because it is small, measuring just over 3g and measuring about 25 × 24 × 9 mm, the camera is shown in Figure 3.…”