Forensic Gait Analysis
Forensic Gait Analysis
Embedded Files
Biometric identification refers to the process of recognizing individuals based on their unique physiological or behavioral characteristics.
Among various biometric signals, gait possesses distinctive advantages over other biometric modalities
in that it can be acquired at a distance and captured using conventional video devices such as CCTV cameras.
Our laboratory investigates methods for analyzing discriminative gait dynamics from video- or sensor-based gait data and
applying them to person re-identification tasks.
On Going Works
Moving Montage: Moving Montage: AI-based
Person Re-Identification from 3D Gait Sequences
Reconstructed from 2D Videos
Reconstructs 3D gait motion of pedestrians from gait videos captured by surveillance
systems such as CCTV cameras.Identifies individuals by analyzing discriminative 3D gait dynamics that are difficult to
preserve using only a single 2D video.Develops a robust AI-based biometric identification system that maintains reliable performance under varying gait conditions and is suitable for potential use as forensic evidence.
Related Publications
HAMGait: A Hybrid Attention-Mamba Network for Cross-View Gait Recognition
Hyebin Kim, Junhyug Noh, Taeyong Lee
GaitGraph3D: A 3D Skeleton-Based Gait Recognition Framework with Segment-Level and Metric Learning Analysis
Hyebin Kim, Hyesu Son, Taeyong Lee
Page updated
Report abuse