Artificial Intelligence (AI) can identify people by the way they walk. The technology focuses on how a person’s joints move, rather than on body shape alone, and could improve long-distance identity verification for security and law enforcement.
Writing in the International Journal of Reasoning-based Intelligent Systems, the team describes the SKDMap-Net system, which analyses a person’s gait using estimated body keypoints from video input. The system calculates joint positions, angles, and angular velocity and acceleration to capture the distinctive features of an individual’s gait. It copes with the effects of different types of clothing, camera angle, and even partial obstruction.
The model processes body position and movement information separately before combining them. It also uses an attention mechanism, a machine learning technique that assigns greater importance to different body parts depending on the scene, such as arm movements if the legs are obscured.
In tests on three public gait-recognition datasets, the system outperformed existing approaches. The approach could make gait recognition more reliable and at the same time reduce the amount of personal visual information that must be processed.
Quan, B. and Zhang, B. (2026) ‘Dual-stream spatiotemporal fusion with dynamic feature mapping for gait-based identity recognition’, Int. J. Reasoning-based Intelligent Systems, Vol. 18, No. 17, pp.67–82.
News media may use this press release as source material, in whole or in part, provided the content is not materially misrepresented. A link back to the original article is appreciated.
No comments:
Post a Comment