Deep Neural Networks (DNNs) for Images Analysis
Mehdi Ghayoumi · 2021
Deep learning (DL) algorithms have shown promising results in many applications using different data types, from images to stock market data. This chapter reviews how to deploy DL’s methods for image analysis, and discusses image analysis, object recognition, image classification, image segmentation, and generation projects using DL’s algorithms. It explains some critical concepts, such as convolution, pooling, and finding patterns in the images. The DL methods for image analysis applications have two main types supervised and unsupervised learning. Stride is the value that filters move over the image in each processing step. The chapter presents some of the most popular convolutional neural networks (CNN) algorithms. Object recognition is one of the best examples of using CNN for a real-life problem. One application of CNN is in image classification. The chapter explains image classification steps using CNN, with example. Image segmentation is partitioning the segments that are meaningful and help to analyze the images better and easier.