Interactive co-segmentation with projection learning

• Grant number: 19/21734-9
• Support type: Scholarships abroad - Research Internship - Scientific Initiation
• Effective date (Start): December 01, 2019
• Effective date (End): March 14, 2020
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal researcher: Alexandre Xavier Falcão
• Grantee: Jordão Okuma Barbosa Ferraz Bragantini
• Supervisor abroad: Laurent Najman
• Home Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Research place: École Supérieure d'Ingénieurs en Électrotechnique et Électronique (ESIEE), France
• Associated to the scholarship: 19/11349-0 - Image segmentation based on dynamic trees and neural networks, BP.IC

Abstract

Image segmentation is one of the most studied topics in the computer vision community today. In this context, weakly supervised methods are starting to gain attention because of the vast amount of labeled data necessary for algorithms with standard training procedure. Another approach to tackle this problem is to improve the existing methods for image annotation, such as interactive image segmentation methods, facilitating data annotation. Therefore,in this project, the goal is to combine the existing graph-based image segmentation methods, researched on the student previous project, FAPESP Project2018/08951-8, with broader modeling techniques to perform multiple images segmentation with minimal interaction, while enabling user interaction outside of the image domain, in this case, in the projection space. (AU)