An Introduction to Image Classification: From Designed Models to End-to-End Learning

Image classification is a crucial component of computer vision tasks, and traditional methods involve feature extraction and classification in feature space. Current state-of-the-art methods utilize end-to-end learning with deep neural networks, where feature extraction and classification are integrated into the model. This book provides a comprehensive introduction to image classification, covering model-driven feature extraction, probabilistic classification, neural networks, regularization, and analyzing learned models. Practical exercises in Python/Keras/Tensorflow are designed to allow students to experiment with these concepts.

Format: Hardback
Length: 290 pages
Publication date: 25 January 2024
Publisher: Springer Verlag, Singapore

Image classification is a crucial aspect of computer vision tasks, with numerous applications across various domains. Traditional methods for image classification involve the extraction of features and their classification in feature space. However, in recent years, state-of-the-art methods have shifted towards end-to-end learning with deep neural networks, where feature extraction and classification are seamlessly integrated into the model.

Understanding the traditional image classification process is essential as it sheds light on the underlying concepts that are fundamental to neural networks. Many of the design principles employed in traditional image classification are directly related to the components of a neural network. This knowledge can help demystify the behavior of these networks, which may initially appear opaque.

The book begins by introducing methods for model-driven feature extraction and classification, encompassing fundamental computer vision techniques for extracting high-level semantics from images. A brief overview of probabilistic classification, encompassing generative and discriminative classifiers, is then provided.

Neural networks are then presented as a powerful tool for learning a classification model directly from labeled sample images. The book delves into the individual components of a neural network, exploring their relationships with those of a traditional designed model. Different concepts for regularizing model training are explained, ensuring that the models are optimized and perform well.

The closing section of the book focuses on various methods for analyzing what a network has learned. This includes techniques such as visualizing the learned features, evaluating the performance of the classifier, and understanding the decision-making process of the network.

Throughout the book, practical exercises in Python/Keras/Tensorflow have been designed to allow students to experiment with these concepts hands-on. Each chapter introduces suitable functions from Python modules to provide students with a solid foundation for understanding and applying image classification techniques.

In summary, image classification is a critical component of computer vision, and traditional and modern methods have evolved significantly. Understanding the traditional image classification process is essential for appreciating the underlying concepts of neural networks. By exploring these concepts through practical exercises and theoretical discussions, students can gain a deep understanding of image classification and its applications in various fields.

Weight: 619g
Dimension: 235 x 155 (mm)
ISBN-13: 9789819978816
Edition number: 1st ed. 2024