The Pattern Recognition and Earth Observation (PATREO) Laboratory, where the following research is currently taking place, accomodates dozens of students from technical high school students to PhD Candidates, researching on several fields such us Remote Sensing, Content-Based Image Retrieval, Digital Forensics Analysis, and Geographic Information Systems [source].
During my PhD, I intend to tackle the problem of CSAM Detection from an ethical point of view. Due to the sensitive nature of this research, it is important to account for the rights of children and adolescents, not only the ones victimized by criminals, but the ones represented on datasets used for training machine learning models. We intend to propose a technique that minimizes harm to children and maximizes benefits for society as a whole.
The project is a partnership with Brazil's Federal Police.
Laranjeira da Silva, Camila, et al. "Seeing without looking: Analysis pipeline for child sexual abuse datasets." 2022 ACM Conference on Fairness, Accountability, and Transparency. 2022.
Specialized technical consultancy for research and development in the area of Applied Artificial Intelligence and Big Data solutions, linked to the scope of the Public Ministry of Minas Gerais (MPMG) Analytical Capabilities Program, including knowledge transfer, through mentoring and assisted operation. During that project, we developed an application for the detection of CSAM in images and videos. The application was based on the work of [1], previously developed at PATREO, which was tested against real CSAM data.
[1] Macedo, Joao, Filipe Costa, and Jefersson A. dos Santos. "A benchmark methodology for child pornography detection." 2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI). IEEE, 2018.