This week at Computer Vision Talks, we host Krishna Chaitanya, PhD Student
Computer Vision Lab, ETH Zurich to talk about his work ‘Contrastive learning of global and local features for medical image segmentation with limited annotations’ accepted at NeurIPS 2020.
What does the talk cover?
- How to achieve high segmentation performance for medical imaging tasks with limited annotations?
- Discuss methods in literature to address limited annotations problem.
- Can we use unlabeled images to pre-train the network using self-supervised learning (SSL) methods?
- Idea: use Contrastive learning (SSL method) for this pre-training.
Read his paper - [2006.10511] Contrastive learning of global and local features for medical image segmentation with limited annotations
Join us at 10:30AM CET this Saturday, 23rd January 2020 - Meeting Registration - Zoom