YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition
Soft biometrics inference in surveillance scenarios is a topic of interest for various applications, particularly in security-related areas. However, soft biometric analysis is not extensively reported in <italic>wild conditions</italic>. In particular, previous works on gender recogniti...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9730882/ |
| _version_ | 1811329538862350336 |
|---|---|
| author | Tiago Roxo Hugo Proenca |
| author_facet | Tiago Roxo Hugo Proenca |
| author_sort | Tiago Roxo |
| collection | DOAJ |
| description | Soft biometrics inference in surveillance scenarios is a topic of interest for various applications, particularly in security-related areas. However, soft biometric analysis is not extensively reported in <italic>wild conditions</italic>. In particular, previous works on gender recognition report their results in face datasets, with relatively good image quality and frontal poses. Given the uncertainty of the availability of the facial region in wild conditions, we consider that these methods are not adequate for surveillance settings. To overcome these limitations, we: 1) present frontal and <italic>wild</italic> face versions of three well-known surveillance datasets; and 2) propose YinYang-Net (YY-Net), a model that effectively and dynamically complements facial and body information, which makes it suitable for gender recognition in wild conditions. The frontal and <italic>wild</italic> face datasets derive from widely used Pedestrian Attribute Recognition (PAR) sets (PETA, PA-100K, and RAP), using a pose-based approach to filter the frontal samples and facial regions. This approach retrieves the facial region of images with varying image/subject conditions, where the state-of-the-art face detectors often fail. YY-Net combines facial and body information through a learnable fusion matrix and a channel-attention sub-network, focusing on the most influential body parts according to the specific image/subject features. We compare it with five PAR methods, consistently obtaining state-of-the-art results on gender recognition, and reducing the prediction errors by up to 24% in frontal samples. The announced PAR datasets versions and YY-Net serve as the basis for <italic>wild</italic> soft biometrics classification and are available in here. |
| first_indexed | 2024-04-13T15:45:40Z |
| format | Article |
| id | doaj.art-ad938a3172fc434e94b52dc8d6fc2cde |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-13T15:45:40Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-ad938a3172fc434e94b52dc8d6fc2cde2022-12-22T02:40:59ZengIEEEIEEE Access2169-35362022-01-0110281222813210.1109/ACCESS.2022.31578579730882YinYang-Net: Complementing Face and Body Information for Wild Gender RecognitionTiago Roxo0https://orcid.org/0000-0001-9563-8039Hugo Proenca1https://orcid.org/0000-0003-2551-8570Instituto de TelecomunicaÇões (IT), University of Beira Interior, Covilhã, PortugalInstituto de TelecomunicaÇões (IT), University of Beira Interior, Covilhã, PortugalSoft biometrics inference in surveillance scenarios is a topic of interest for various applications, particularly in security-related areas. However, soft biometric analysis is not extensively reported in <italic>wild conditions</italic>. In particular, previous works on gender recognition report their results in face datasets, with relatively good image quality and frontal poses. Given the uncertainty of the availability of the facial region in wild conditions, we consider that these methods are not adequate for surveillance settings. To overcome these limitations, we: 1) present frontal and <italic>wild</italic> face versions of three well-known surveillance datasets; and 2) propose YinYang-Net (YY-Net), a model that effectively and dynamically complements facial and body information, which makes it suitable for gender recognition in wild conditions. The frontal and <italic>wild</italic> face datasets derive from widely used Pedestrian Attribute Recognition (PAR) sets (PETA, PA-100K, and RAP), using a pose-based approach to filter the frontal samples and facial regions. This approach retrieves the facial region of images with varying image/subject conditions, where the state-of-the-art face detectors often fail. YY-Net combines facial and body information through a learnable fusion matrix and a channel-attention sub-network, focusing on the most influential body parts according to the specific image/subject features. We compare it with five PAR methods, consistently obtaining state-of-the-art results on gender recognition, and reducing the prediction errors by up to 24% in frontal samples. The announced PAR datasets versions and YY-Net serve as the basis for <italic>wild</italic> soft biometrics classification and are available in here.https://ieeexplore.ieee.org/document/9730882/Gender recognitionselective attentionsoft biometrics analysisvisual surveillance |
| spellingShingle | Tiago Roxo Hugo Proenca YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition IEEE Access Gender recognition selective attention soft biometrics analysis visual surveillance |
| title | YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition |
| title_full | YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition |
| title_fullStr | YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition |
| title_full_unstemmed | YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition |
| title_short | YinYang-Net: Complementing Face and Body Information for Wild Gender Recognition |
| title_sort | yinyang net complementing face and body information for wild gender recognition |
| topic | Gender recognition selective attention soft biometrics analysis visual surveillance |
| url | https://ieeexplore.ieee.org/document/9730882/ |
| work_keys_str_mv | AT tiagoroxo yinyangnetcomplementingfaceandbodyinformationforwildgenderrecognition AT hugoproenca yinyangnetcomplementingfaceandbodyinformationforwildgenderrecognition |