This book covers aspects of human re-identification problems related to computer vision and machine learning. Working from a practical perspective, it introduces novel algorithms and designs for human re-identification that bridge the gap between research and reality. The primary focus is on building a robust, reliable, distributed and scalable smart surveillance system that can be deployed in real-world scenarios. This book also includes detailed discussions on pedestrian candidates detection, discriminative feature extraction and selection, dimension reduction, distance/metric learning, and decision/ranking enhancement.This book is intended for professionals and researchers working in computer vision and machine learning. Advanced-level students of computer science will also find the content valuable.

Covers every aspect of an end-to-end real-world human re-identification system Analyzes and summarizes factors of challenges, risks and uncertainties from practical computer vision applications Extensive evaluation and benchmarking on mainstream human re-identification algorithms and datasets

Inhalt
The Problem of Human re-identification.- Features and Signatures.- Multi-Object Tracking.- Surveillance Camera and its Calibration.- Calibrating a Surveillance Camera Network.- Learning Viewpoint Invariant Signatures.- Learning Subject-Discriminative Features.- Dimension Reduction with Random Projections.- Sample Selection for Multi-shot Human Reidentification.- Conclusions and Future Work.