Алгоритм реидентификации по биометрическому шаблону лица человека
| Parent link: | Перспективы развития фундаментальных наук=Prospects of Fundamental Sciences Development: сборник научных трудов XXI Международной конференции студентов, аспирантов и молодых ученых, г. Томск, 23-26 апреля 2024 г./ Национальный исследовательский Томский политехнический университет ; под ред. И. А. Курзиной [и др.].— .— Томск: Изд-во ТПУ Т. 3 : Математика.— 2024.— С. 5-7 |
|---|---|
| Main Author: | |
| Corporate Author: | |
| Other Authors: | , |
| Summary: | Заглавие с экрана This article provided performance research of DBSCAN and agglomerative clustering algorithms on re-identification task based on face embeddings. Evaluation of algorithms performance was made using custom dataset containing 35 090 images of 20 people. Adjusted mutual information, Purity and Inverse Purity scores are equal to 0.9964, 1.0 and 0.9965 respectively for both algorithms using cosine distance threshold between samples of same cluster equal to 0.37 Текстовый файл |
| Language: | Russian |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://earchive.tpu.ru/handle/11683/80537 |
| Format: | Electronic Book Chapter |
| KOHA link: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=674310 |
| Summary: | Заглавие с экрана This article provided performance research of DBSCAN and agglomerative clustering algorithms on re-identification task based on face embeddings. Evaluation of algorithms performance was made using custom dataset containing 35 090 images of 20 people. Adjusted mutual information, Purity and Inverse Purity scores are equal to 0.9964, 1.0 and 0.9965 respectively for both algorithms using cosine distance threshold between samples of same cluster equal to 0.37 Текстовый файл |
|---|