Subject Identification Based on Gait Using a RGB-D Camera

• Type: Article in International Conference Proceedings Book
• Date: 2020

Authors

Ana Patrícia Rocha (Author) Other; José Maria Fernandes (Author) Other; Hugo Miguel Pereira Choupina (Author) Other; Maria do Carmo Vilas-Boas (Author) Other; João Paulo Silva Cunha (Author) FEUP

Conference proceedings International

Title: Advances in Intelligent Systems and Computing

Pages: 76-85

10th International Conference on Soft Computing and Pattern Recognition, SoCPaR 2018

13 December 2018 through 15 December 2018

Indexing

Scopus - 0 Citations

INSPEC

Scientific classification

CORDIS: Technological sciences > Engineering > Electrical engineering

Other information

Authenticus ID: P-00Q-GF4

DOI: 10.1007/978-3-030-17065-3_8

Abstract (EN): Biometric authentication (i.e., verification of a given subject’s identity using biological characteristics) relying on gait characteristics obtained in a non-intrusive way can be very useful in the area of security, for smart surveillance and access control. In this contribution, we investigated the possibility of carrying out subject identification based on a predictive model built using machine learning techniques, and features extracted from 3-D body joint data provided by a single low-cost RGB-D camera (Microsoft Kinect v2). We obtained a dataset including 400 gait cycles from 20 healthy subjects, and 25 anthropometric measures and gait parameters per gait cycle. Different machine learning algorithms were explored: k-nearest neighbors, decision tree, random forest, support vector machines, multilayer perceptron, and multilayer perceptron ensemble. The algorithm that led to the model with best trade-off between the considered evaluation metrics was the random forest: overall accuracy of 99%, class accuracy of 100±Â0%, and F1 score of 99±Â2%. These results show the potential of using a RGB-D camera for subject identification based on quantitative gait analysis. © 2020, Springer Nature Switzerland AG.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 10