Related Experiment Video
Updated: Jul 17, 2026
06:37
Artificial Intelligence-Based System for Detecting Attention Levels in Students
Published on: December 15, 2023
Personal recognition rate improvement using head-top image and an application to walking subject
K Nakajima1, A Kamiya, K Sasaki
1Division of Bio-Information Engineering, Faculty of Engineering, Toyama University, 3190 Gofuku, Toyama 930-8555, Japan.
Summary
This study enhanced head-top imaging for smart house identification, achieving 100% recognition for stationary subjects. Walking recognition yielded 52% accuracy, indicating potential for dynamic personal identification systems.
Area of Science:
- Computer Vision
- Biometrics
- Smart Home Technology
Background:
- Personal identification is crucial for smart home security and automation.
- Head-top imaging offers a non-intrusive biometric modality.
- Previous methods achieved 86.4% recognition for stationary subjects.
Purpose of the Study:
- To improve the personal identification accuracy using head-top images.
- To evaluate the performance of head-top imaging under dynamic (walking) conditions.
Main Methods:
- Utilized an existing head-top image database of 11 stationary subjects.
- Acquired head-top image sequences of 6 subjects using a thermal camera while walking.
- Applied image processing and recognition algorithms to both static and dynamic datasets.
Main Results:
- Achieved a 100% personal recognition rate for stationary subjects, an improvement from 86.4%.
- Obtained a 52% personal recognition rate for subjects captured while walking.
- Demonstrated the feasibility of head-top imaging for personal identification.
Conclusions:
- Head-top imaging can achieve perfect recognition for stationary individuals in smart home environments.
- Dynamic recognition using head-top images presents challenges but shows potential for real-world applications.
- Further research is needed to enhance recognition rates for moving subjects.
