This chapter provides a comprehensive analysis of methodologies employed in gait identification research, drawing insights from articles gathered in a systematic review spanning 2019–2023. The methodologies are categorized based on their approach: Deep Learning (DL) based and non-Deep Learning (non-DL) based. Traditional non-DL-based methods exhibit limitations in handling large datasets and complex variations. The DL-based methods, leveraging advancements in DL algorithms, demonstrate the capability to extract features automatically, reducing the need for manual feature engineering. Given the predominance of DL-based approaches, a further distinction is made between model-based and model-free approaches. Within DL, model-free approaches are dominant, showcasing a variety of architectures like CNNs, GANs, DAEs, CapsNets, RNNs, and ViTs. These approaches demonstrate strengths in image-based tasks, data generation, unsupervised learning, and sequence modeling. However, model-based approaches have grown recently, particularly with advancements in human pose estimation algorithms. This chapter compares these diverse methodologies, highlighting their strengths, limitations, and contributions to advancing gait identification research.