<p>Person Re-Identification (Re-ID) is a key approach for identifying persons in multiple cameras and frames of images, with widespread applications including surveillance, security, and personalized shopping. We discuss the main application domains of Re-ID, which are: surveillance &amp; security, retail and marketing, smart cities, healthcare, and education, to illustrate how Re-ID systems are reshaping these fields. This paper provides a comprehensive survey of progress in Re-ID, focusing on deep learning approaches, including Convolutional Neural Networks (CNNs), attention mechanisms, Deep Metric Learning (DML), Part-based models, Vision Transformers (ViTs), and Deep Adversarial Learning. The paper presents the main challenges that hinder accurate person identification across different real-world environments: pose variation, illumination variation, occlusion, cross-camera variability, scalability, and identity discrimination. Besides, it summarizes the most relevant datasets and some evaluation metrics that have already worked as benchmarks for the Re-ID models. This survey summarizes existing studies on recent works of Re-ID technology, discussing prominent trends and emerging methods. It also points out the future directions that can be taken. It will help researchers, practitioners, and decision-makers develop the Re-ID systems and solve related problems.</p>