Person Identification by Footstep Sound Using Convolutional Neural Networks
Human gait is very individual and it may serve as biometric to identify people in camera recordings. Comparable results can be achieved while using the acoustic signature of human footstep sounds. This acoustic solution offers the opportunity of less installation space and the use of cost-efficient...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-05-01
|
| Series: | Applied Mechanics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2673-3161/2/2/16 |
| _version_ | 1797534543722315776 |
|---|---|
| author | Stephan Algermissen Max Hörnlein |
| author_facet | Stephan Algermissen Max Hörnlein |
| author_sort | Stephan Algermissen |
| collection | DOAJ |
| description | Human gait is very individual and it may serve as biometric to identify people in camera recordings. Comparable results can be achieved while using the acoustic signature of human footstep sounds. This acoustic solution offers the opportunity of less installation space and the use of cost-efficient microphones when compared to visual system. In this paper, a method for person identification based on footstep sounds is proposed. First, step sounds are isolated from microphone recordings and separated into 500 ms samples. The samples are transformed with a sliding window into mel-frequency cepstral coefficients (MFCC). The result is represented as an image that serves as input to a convolutional neural network (CNN). The dataset for training and validating the CNN is recorded with five subjects in the acoustic lab of DLR. These experiments identify a total number of 1125 steps. The validation of the CNN reveals a minimum <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>F</mi><mn>1</mn></msub></semantics></math></inline-formula>-score of 0.94 for all five classes and an accuracy of 0.98. The Grad-CAM method is applied to visualize the background of its decision in order to verify the functionality of the proposed CNN. Subsequently, two challenges for practical implementations, noise and different footwear, are discussed using experimental data. |
| first_indexed | 2024-03-10T11:30:59Z |
| format | Article |
| id | doaj.art-ba4e4ee6656d4bb996bf830274f6a37d |
| institution | Directory Open Access Journal |
| issn | 2673-3161 |
| language | English |
| last_indexed | 2024-03-10T11:30:59Z |
| publishDate | 2021-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Mechanics |
| spelling | doaj.art-ba4e4ee6656d4bb996bf830274f6a37d2023-11-21T19:15:02ZengMDPI AGApplied Mechanics2673-31612021-05-012225727310.3390/applmech2020016Person Identification by Footstep Sound Using Convolutional Neural NetworksStephan Algermissen0Max Hörnlein1German Aerospace Center (DLR), Lilienthalplatz 7, 38108 Braunschweig, GermanyGerman Aerospace Center (DLR), Lilienthalplatz 7, 38108 Braunschweig, GermanyHuman gait is very individual and it may serve as biometric to identify people in camera recordings. Comparable results can be achieved while using the acoustic signature of human footstep sounds. This acoustic solution offers the opportunity of less installation space and the use of cost-efficient microphones when compared to visual system. In this paper, a method for person identification based on footstep sounds is proposed. First, step sounds are isolated from microphone recordings and separated into 500 ms samples. The samples are transformed with a sliding window into mel-frequency cepstral coefficients (MFCC). The result is represented as an image that serves as input to a convolutional neural network (CNN). The dataset for training and validating the CNN is recorded with five subjects in the acoustic lab of DLR. These experiments identify a total number of 1125 steps. The validation of the CNN reveals a minimum <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mi>F</mi><mn>1</mn></msub></semantics></math></inline-formula>-score of 0.94 for all five classes and an accuracy of 0.98. The Grad-CAM method is applied to visualize the background of its decision in order to verify the functionality of the proposed CNN. Subsequently, two challenges for practical implementations, noise and different footwear, are discussed using experimental data.https://www.mdpi.com/2673-3161/2/2/16person identificationconvolutional neural networksMFCCgait recognitionmachine learning |
| spellingShingle | Stephan Algermissen Max Hörnlein Person Identification by Footstep Sound Using Convolutional Neural Networks Applied Mechanics person identification convolutional neural networks MFCC gait recognition machine learning |
| title | Person Identification by Footstep Sound Using Convolutional Neural Networks |
| title_full | Person Identification by Footstep Sound Using Convolutional Neural Networks |
| title_fullStr | Person Identification by Footstep Sound Using Convolutional Neural Networks |
| title_full_unstemmed | Person Identification by Footstep Sound Using Convolutional Neural Networks |
| title_short | Person Identification by Footstep Sound Using Convolutional Neural Networks |
| title_sort | person identification by footstep sound using convolutional neural networks |
| topic | person identification convolutional neural networks MFCC gait recognition machine learning |
| url | https://www.mdpi.com/2673-3161/2/2/16 |
| work_keys_str_mv | AT stephanalgermissen personidentificationbyfootstepsoundusingconvolutionalneuralnetworks AT maxhornlein personidentificationbyfootstepsoundusingconvolutionalneuralnetworks |