Yannis Kalantidis

Papers

Total Citations

H-Index

About

Yannis Kalantidis is a leading researcher in computer vision and machine learning, with a primary focus on visual representation learning, large-scale retrieval, and human pose modeling. His most influential work includes pioneering contributions to unsupervised and self-supervised learning, particularly through methods like "SwAV" (Swapping Assignments between Views), which revolutionized how models learn visual features without labels by leveraging online clustering and contrastive learning. This work has amassed thousands of citations and is considered foundational in modern self-supervised learning. Kalantidis has also made significant strides in temporal human modeling, as evidenced by his work on "PoseBERT," a generic transformer module for 3D human pose estimation in videos, which addresses the challenge of expensive annotation by enabling state-of-the-art performance with limited labeled data. Beyond his research, he has served as a research scientist at Meta (Facebook AI Research) and NAVER LABS Europe, where his innovations have directly influenced industry-scale visual search and understanding systems. His work consistently bridges theoretical advances with practical, scalable solutions, making him a highly cited and respected figure in the computer vision community.