ReLISA:Reinforcement Learning Interactive Segmentation Annotator Agent Integrated in an Instance Segmentation Workflow

Nowadays, with the advent and increasing importance of Deep Learning and Machine Learning techniques, the process of labeling data is becoming more and more important for several problems. In particular for computer vision segmentation tasks, having the possibility to create precise mask annotations of images in a reasonable amount of time is crucial. For this reason, it was decided to use deep learning and deep reinforcement learning techniques in order to propose a user-friendly platform that provides the possibility to create precise masks of specific targets in an image manually and automatically through the introduction of a Reinforcement Learning Interactive Segmentation Annotator (ReLISA) agent. This thesis will present these solutions and their application in an overall instance segmentation workflow pointing out possible applications and future improvements.