Unsupervised Domain Adaptation Approaches for Person Localization in the OR

The fine-grained localization of clinicians in the operating room (OR) is a key component in designing the new OR support systems. However, the task is challenging not only because OR images contain significant visual domain differences compared to traditional vision datasets but also because data and annotations are hard to collect and generate in the OR due to privacy concerns. In this talk, we will explore Unsupervised Domain Adaptation (UDA) methods to enable visual learning for the target domain, the OR, by working in two complementary directions. First, we study how low-resolution images with a downsampling factor as low as 12x can be used for fine-grained clinicians' localization to address privacy concerns. Second, we propose several self-supervised methods to transfer learned information from a labeled source domain to an unlabeled target domain to address the shift of visual domain and lack of annotations. These methods employ self-supervised predictions in allowing the model to learn and adapt to the unlabeled target domain.

Speaker: Vinkle Kumar, ICube lab, University of Strasbourg