This paper conducts a systematic comparative analysis of marker-based and markerless gait analysis systems, focusing on experimental methodologies, hardware architecture, artificial intelligence integration, and information systems. Performance is evaluated through accuracy, reliability, and validity metrics in spatiotemporal analysis, kinematic assessment, and kinetic assessment. Empirical data from over 20 original studies demonstrate that marker-based systems excel in kinematic precision (root mean square error < 3°), while markerless systems, though clinically acceptable in accuracy (joint angle error < 5°), offer superior scalability and ecological validity. This paper elucidates the key differences and synergistic advantages of both approaches, advocating for hybrid solutions to enhance diagnostic precision and enable personalized interventions in AI-enhanced biomechanics.