Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System
Person Re-Identification is an essential task in computer vision, particularly in surveillance applications. The aim is to identify a person based on an input image from surveillance photographs in various scenarios. Most Person re-ID techniques utilize Convolutional Neural Networks (CNNs); however,...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-10-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/11/19/9197 |
| _version_ | 1827680919094296576 |
|---|---|
| author | Muhammad Tahir Saeed Anwar |
| author_facet | Muhammad Tahir Saeed Anwar |
| author_sort | Muhammad Tahir |
| collection | DOAJ |
| description | Person Re-Identification is an essential task in computer vision, particularly in surveillance applications. The aim is to identify a person based on an input image from surveillance photographs in various scenarios. Most Person re-ID techniques utilize Convolutional Neural Networks (CNNs); however, Vision Transformers are replacing pure CNNs for various computer vision tasks such as object recognition, classification, etc. The vision transformers contain information about local regions of the image. The current techniques take this advantage to improve the accuracy of the tasks underhand. We propose to use the vision transformers in conjunction with vanilla CNN models to investigate the true strength of transformers in person re-identification. We employ three backbones with different combinations of vision transformers on two benchmark datasets. The overall performance of the backbones increased, showing the importance of vision transformers. We provide ablation studies and show the importance of various components of the vision transformers in re-identification tasks. |
| first_indexed | 2024-03-10T07:06:07Z |
| format | Article |
| id | doaj.art-cff1b127d7aa44269614d5afd6c1be3d |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T07:06:07Z |
| publishDate | 2021-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-cff1b127d7aa44269614d5afd6c1be3d2023-11-22T15:49:10ZengMDPI AGApplied Sciences2076-34172021-10-011119919710.3390/app11199197Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance SystemMuhammad Tahir0Saeed Anwar1College of Computing and Informatics, Saudi Electronic University, Riyadh 11673, Saudi ArabiaData61-Commonwealth Scientific and Industrial Research Organization(CSIRO), Clayton South, VIC 3169, AustraliaPerson Re-Identification is an essential task in computer vision, particularly in surveillance applications. The aim is to identify a person based on an input image from surveillance photographs in various scenarios. Most Person re-ID techniques utilize Convolutional Neural Networks (CNNs); however, Vision Transformers are replacing pure CNNs for various computer vision tasks such as object recognition, classification, etc. The vision transformers contain information about local regions of the image. The current techniques take this advantage to improve the accuracy of the tasks underhand. We propose to use the vision transformers in conjunction with vanilla CNN models to investigate the true strength of transformers in person re-identification. We employ three backbones with different combinations of vision transformers on two benchmark datasets. The overall performance of the backbones increased, showing the importance of vision transformers. We provide ablation studies and show the importance of various components of the vision transformers in re-identification tasks.https://www.mdpi.com/2076-3417/11/19/9197vision transformersdeep learningre-IDimage retrievalmulti-camera surveillance systempedestrian identification |
| spellingShingle | Muhammad Tahir Saeed Anwar Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System Applied Sciences vision transformers deep learning re-ID image retrieval multi-camera surveillance system pedestrian identification |
| title | Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System |
| title_full | Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System |
| title_fullStr | Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System |
| title_full_unstemmed | Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System |
| title_short | Transformers in Pedestrian Image Retrieval and Person Re-Identification in a Multi-Camera Surveillance System |
| title_sort | transformers in pedestrian image retrieval and person re identification in a multi camera surveillance system |
| topic | vision transformers deep learning re-ID image retrieval multi-camera surveillance system pedestrian identification |
| url | https://www.mdpi.com/2076-3417/11/19/9197 |
| work_keys_str_mv | AT muhammadtahir transformersinpedestrianimageretrievalandpersonreidentificationinamulticamerasurveillancesystem AT saeedanwar transformersinpedestrianimageretrievalandpersonreidentificationinamulticamerasurveillancesystem |