개인 인증을 위한 CNN 기반 보행 패턴 식별 = Identification of Gait Patterns using Convolutional Neural Networks for Personal Authentication

다국어 초록 (Multilingual Abstract)

Efforts are underway to utilize the unique characteristics of human gait as a tool for user authentication in the latest gait analysis study. In this paper, we study the identification of pedestrians by learning human gait data acquired by an inertial sensor with a CNN model. The model proposed in this paper is evaluated using accuracy, precision, recall, F1-score, ROC Curve, AUC, which are used as general evaluation indicators, and FAR, FRR, and EER, which are biometric evaluation matrices. As a result, the accuracy, precision, recall, and F1-score are all over 99%. In addition, the average AUC is 0.99, and the average EER is 0.0009, which leads to the conclusion that the model proposed in the study can identify human gait with high reliability and can be used as a method for user authentication.

국문 초록 (Abstract)

최신의 보행 분석 연구에서 사람의 걸음걸이가 지닌 고유한 특징을 사용자 인증 수단으로 활용하려는 노력이 진행 중이다. 본 논문에서는 관성 센서로 획득한 인간의 걸음걸이 데이터를 CNN 모델로 학습하여 보행자가 누구인지 식별하는 것을 연구한다. 학습 모델의 평가지표로 흔히 사용하는 정확도, 정밀도, 재현율, F1-score, ROC Curve, AUC와 생체인증 평가지표인 FAR, FRR, EER을 활용하여 본 논문에서 제안한 모델을 평가하였다. 그 결과, 정확도, 정밀도, 재현율, F1-score 모두 99% 이상의 결과를 얻었다. 또한, 평균 AUC는 0.99, 평균 EER은 0.0009로 연구에서 제안된 모델이 높은 신뢰성을 가지고 사람의 걸음걸이를 식별할 수 있으며 사용자 인증을 위한 수단으로 사용 가능하다는 결론에 도달했다.

참고문헌 (Reference)

1 G. Giorgi, "Using recurrent neural networks for continuous authentication through gait analysis" 147 : 157-163, 2021

2 Y. Li, "Using Data Augmentation in Continuous Authentication on Smartphones" 6 (6): 628-640, 2019

3 L. V. R. Asuncion, "Thigh Motion-Based Gait Analysis for Human Identification using Inertial Measurement Units (IMUs)" 1-6, 2018

4 M. Muaaz, "Smartphone-Based Gait Recognition: From Authentication to Imitation" 16 (16): 3209-3221, 2017

5 L. Tran, "Multi-Model Long Short-Term Memory Network for Gait Recognition Using Window-Based Data Segment" 9 : 23826-23839, 2021

6 최지우 ; 유형진 ; 최상일 ; 강태원, "LSTM을 사용한 보행주기 식별" 한국정보기술학회 19 (19): 27-39, 2021

7 W. Xu, "KEH-Gait: Using Kinetic Energy Harvesting for Gait-based User Authentication Systems" 18 (18): 139-152, 2019

8 O. Adel, "Inertial Gait-based Person Authentication Using Siamese Networks" 1-7, 2021

9 R. Luca, "Inertial Data Based Learning Methods for Person Authentication" 1-4, 2021

10 M. Gadaleta, "IDNet: Smartphone-ba sed gait recognition with convolutional neural networks" 74 : 25-37, 2018

11 R. San-Segundo, "I-vector analysis for Gait-based Person Identification using smartphone inertial signals" 38 (38): 140-153, 2017

12 H. Chunsheng, "Human Gait Feature Data Analysis and Person Identification Based on IMU" 37-442, 2020

13 I. Papavasileiou, "GaitCode: Gait-based continuous authentication using multimodallearning and wearable sensors" 19 : 100162-, 2021

14 W. Xu, "Gait-Watch: A Gait-based contextaware authentication system for smart watch via sparse coding" 107 : 102218-, 2020

15 Y. Watanabe, "Gait identification and authentication using LSTM based on 3-axis accelerations of smartphone" 176 : 3873-3880, 2020

16 Y. Kim, "Fourth Industrial Revolution(4IR)Hyper-Connected Society and Internet of Things Age" 17 (17): 14-19, 2019

17 Q. Zou, "Deep Learning-Based Gait Recognition Using Smartphones in the Wild" 15 : 3197-3212, 2020

18 M. Ehatisham-ul-Haq, "Continuous authentication of smartphone users based on activity pattern recognit ion using passive mobile sensing" 109 : 24-35, 2018

19 Y. Tsai, "Center-Assisted Personal Gait Authentication Using Orientation Adversarial Feature Extraction" 1-6, 2019

20 R. Delgado-Escaño, "An End-to-End Multi-Task and Fusion CNN for Inertial-Based Gait Recognition" 7 : 1897-1908, 2019

21 S. R. V. Sudhakar, "ActID: An efficient framework for activity sensor based user identification" 108 : 102319-, 2021

22 R. Leyva, "Accelerometer Dense Trajectories for Activity Recognition and People Identification" 1-6, 2019

23 F. Sun, "Accelerometer -Based Speed-Adaptive Gait Authentication Method for Wearable IoT Devices" 6 (6): 820-830, 2019

24 M. Abuhamad, "AUToSen : Deep-Learning-Based Implicit Continuous Authentication Using Smartphone Sensors" 7 (7): 5008-5020, 2020