Lahiru Wijayasingha

Hi, I’m Lahiru. I build machine learning systems that see, hear and sense the world: models for video, images, audio and wearable sensors.

My research centres on computer vision and multimodal deep learning, especially monocular depth estimation, camera-independent 3D perception, and egocentric and wearable sensing. I’m most interested in where these meet people: healthcare, human activity recognition and assistive systems. Recently I’ve been working on explainable AI for video models, finding the smallest set of frames that actually drives a model’s decision.

I did my PhD in Computer Science at the University of Virginia, an MS in Medical Robotics and Image Guided Intervention at Imperial College London, and a BSc in Mechanical Engineering at the University of Moratuwa, Sri Lanka. After that I was a research scholar at the University of Iowa, building multimodal models that combine echocardiogram video with clinical data.

This site is where I keep my CV, projects and papers. It’s also where I write and jot down stray thoughts, usually by the scenic route.