We study learning mid-level representation from natural non-curated data to achieve efficient and generalizing performance on downstream visual tasks such as recognition, segmentation, and detection. We exploit instance discrimination, instance grouping, model bias and variance analysis, pixel-to-segment contrastive learning, and visual memory
ICSI
ICSI is a leading independent, nonprofit center for research in computer science.
With its unique focus on international collaboration and through its established international programs, ICSI brings together scientists from all over the world and at all stages of their career to work with established staff researchers, UC Berkeley professors, and their networks of academic, government, and industrial partners. More >>
RESEARCH AREAS:
Photos found on Flickr, licensed by Creative Commons. See below for credits.
