Pressure-based gait recognition: Feature extraction techniques for footwear-agnostic identity recognition
| dc.contributor.advisor | Scheme, Erik | |
| dc.contributor.author | Salehi, Ala | |
| dc.date.accessioned | 2025-02-19T15:27:44Z | |
| dc.date.available | 2025-02-19T15:27:44Z | |
| dc.date.issued | 2024-12 | |
| dc.description.abstract | This research explores the development of a robust pressure-based gait recognition system, with a focus on reducing the impact of changes in footwear. Using two datasets; CASIA-D and a newly collected UNB dataset, we compare traditional and deep learning methods, including two novel architectures: UMAPNet for spatial feature learning and FootPart, a comprehensive spatiotemporal model. FootPart integrates local spatial partitioning with dynamic temporal modelling, achieving significant improvements in both closed-set and open-set verification tasks. Results show that FootPart maintains high accuracy under variable conditions, outperforming baseline models in identification tasks and demonstrating resilience to unseen footwear. This work underscores the importance of detailed spatial and temporal features in robust gait recognition, with implications for security, healthcare, and smart environments. | |
| dc.description.copyright | © Ala Salehi, 2024 | |
| dc.format.extent | xii, 91 | |
| dc.format.medium | electronic | |
| dc.identifier.uri | https://unbscholar.lib.unb.ca/handle/1882/38252 | |
| dc.language.iso | en | |
| dc.publisher | University of New Brunswick | |
| dc.relation | CyberNB | |
| dc.relation | Knowledge Park | |
| dc.relation | Stepscan Technologies | |
| dc.relation | New Brunswick Innovation Foundation | |
| dc.relation | Atlantic Canada Opportunities Agency (ACOA) | |
| dc.relation | Natural Sciences and Engineering Research Council of Canada (NSERC) | |
| dc.rights | http://purl.org/coar/access_right/c_abf2 | |
| dc.subject.discipline | Electrical and Computer Engineering | |
| dc.title | Pressure-based gait recognition: Feature extraction techniques for footwear-agnostic identity recognition | |
| dc.type | master thesis | |
| oaire.license.condition | other | |
| thesis.degree.discipline | Electrical and Computer Engineering | |
| thesis.degree.grantor | University of New Brunswick | |
| thesis.degree.level | masters | |
| thesis.degree.name | M.Sc.E. |