RT info:eu-repo/semantics/conferenceProceedings
T1 A data augmentation strategy for improving age estimation to support CSEM detection
A1 Chaves, Deisy
A1 Agarwal, Nancy
A1 Fidalgo Fernández, Eduardo
A1 Alegre Gutiérrez, Enrique
A2 Ingenieria de Sistemas y Automatica
K1 Ingeniería de sistemas
K1 Age estimation
K1 Data augmentation
K1 Generative Adversarial Networks
K1 Facial occlusion
K1 CSEM
K1 3310.05 Ingeniería de Procesos
AB [EN] Leveraging image-based age estimation in preventing Child Sexual Exploitation Material (CSEM) contentover the internet is not investigated thoroughly in the research community. While deep learning methodsare considered state-of-the-art for general age estimation, they perform poorly in predicting the age group ofminors and older adults due to the few examples of these age groups in the existing datasets. In this work, wepresent a data augmentation strategy to improve the performance of age estimators trained on imbalanced databased on synthetic image generation and artificial facial occlusion. Facial occlusion is focused on modelling asCSEM criminals tend to cover certain parts of the victim, such as the eyes, to hide their identity. The proposedstrategy is evaluated using the Soft Stagewise Regression Network (SSR-Net), a compact size age estimatorand three publicly available datasets composed mainly of non-occluded images. Therefore, we create theSynthetic Augmented with Occluded Faces (SAOF-15K) dataset to assess the performance of eye and mouthoccludedimages. Results show that our strategy improves the performance of the evaluated age estimator.
PB SciTePress
SN 978-989-758-634-7
LK https://hdl.handle.net/10612/17290
UL https://hdl.handle.net/10612/17290
NO Chaves, D.; Agarwal, N.; Fidalgo, E. and Alegre, E. (2023). A data augmentation strategy for improving age estimation to support CSEM detection. En Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 5. 692-699. DOI: 10.5220/0011719700003417
DS BULERIA. Repositorio Institucional de la Universidad de León
RD 14-jun-2024