Computer Vision News - November 2022

47 AI-enabled Medical Image Analysis best paper award went to Ramin Ebrahim Nakhli and colleagues, from the University of British Columbia, Canada. The paper proposes an approach to automatically identify cells onH&E-stained slides, without the time-consuming step of annotation. The authors propose a contrastive cell representation learning (CCRL) model for cell clustering that enables model training on larger datasets compared to previous methods. See video below. presentations of cutting-edge research on both academia and industry. From attention-based models to contrastive learning and self-supervised learning, the authors presented several approaches to tackle tasks such as data annotation, cell identification or cancer grading. Out of a set of five pre-selected papers, based on the scores given by the reviewers, the workshop participants voted for the three best digital pathology works . The T OF CCV