Cross-domain visual representations under limited supervision and via transfer of learning

• Grant number: 19/16379-5
• Support type: Regular Research Grants
• Duration: January 01, 2020 - March 31, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Cooperation agreement: University of Surrey
• Mobility Program: SPRINT - Projetos de pesquisa - Mobilidade
• Principal researcher: Moacir Antonelli Ponti
• Grantee: Moacir Antonelli Ponti
• Principal researcher abroad: John Collomosse
• Institution abroad: University of Surrey, England
• Home Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 18/22482-0 - Learning features from visual content under limited supervision using multiple domains, AP.R

Abstract

This is a collaborative research activity and network between the University of Surrey and the University of São Paulo. It tries to fill a gap in the research and development of applications in which it is important to learn representations of complex data such as signals, images, and video. In particular, it addresses this area from a perspective of limited supervision, which means learning even with little or no prior knowledge about the available data. Examples of applications include matching visual content in images from different domains such as hand-drawn sketches and digital photographs or art, anomaly detection in signals produced by trajectories of transportation vehicles, and unusual event detection on video surveillance. This will comprise a collaborative research network funded by FAPESP and University of Surrey jointly, leading to research papers, student and research exchange as well as funding with the submission of research grant proposals. The USP team expertise is signal, image and video analysis and statistical learning. The Surrey team complements research at USP in terms of expertise in cross-domain visual learning and anomaly detection. (AU)