About Me

I am a researcher specializing in Computer Vision, Statistical Learning (temporal modeling), and Human Behavior Analysis. Currently, I am a Research Associate at the School of Computing Science, University of Glasgow. Having previously served as a Postdoctoral Research Associate (2021–2025) at the University of Edinburgh. I earned my PhD in Electrical Engineering from Kyoto University in 2020.

My work centers on vision-based analysis of physical behavior, anomaly detection, temporal modelling, and activity recognition.

Technical Skills

Python MATLAB Computer Vision Machine Learning PyTorch CNN / RNN / VAE / LSTM Gaussian Processes Change Point Detection Hidden Markov Models (HMMs) RGB-D & Egocentric Vision

Experience & Education

University of Glasgow

Research Associate | 2026.4 – Present

Developing computer vision and machine learning algorithms for monitoring infantile epileptic spasms.

University of Edinburgh

Visitor & Postdoctoral Research Associate | 2021.5 – 2026.6

Developing computer vision and machine learning algorithms for in-home monitoring of older adults. Modeled physical states using Bayesian Networks, Gaussian Processes, and utilized deep learning (CNNs, LSTM, VAE) and temporal behaviour analysis.

Kyoto University

PhD in Electrical Engineering | 2015.10 – 2020.8

Dissertation: Analysis and Modeling of Machine Operation Tasks Using Egocentric Vision. Developed methods for touch sensing, hotspot detection, and complex task sequence modeling using modified Hidden Markov Models and dynamic alignment.

Sichuan University

Master of Engineering, Control Theory and Engineering | 2011 – 2014

Focused on video denoising, image processing, compressed sensing, and low-rank methods.

Featured Research Projects

Vision-Based Human Physical Behavior Analysis (2021–2025)

Developed privacy-preserving methods for sensing physical behavior in domestic environments. My work utilized RGB-D imaging, activity recognition, behavior analysis, and temporal modeling (Bayesian Networks, Gaussian Processes) to infer changes in behavioral states. Key contributions include handling real-world challenges like low lighting and moving pets to detect anomalies, inactivity, and subtle functional decline reliably.

Human Behavior Analysis Temporal Modeling Gaussian Processes Anomaly Detection Change Point Detection
Temporal modeling plots
Inferred changes in physical activity.

Analysis and Modeling of Machine Operation Tasks using Egocentric Vision (2015–2020)

Investigated operational behaviors using head-mounted cameras, IMUs, and gaze trackers. I developed methods for touch sensing and hotspot detection, and applied Modified Hidden Markov Models to model complex task sequences (repetitive, mistakes, new ways) for automatic guidance systems and user skill assessment.

Egocentric Vision Activity Recognition HMMs
Egocentric vision process
Hotspot detection and complex task sequence modelling.

Selected Publications

Open Source Datasets

Human MotionLess (HuMoLs) Dataset
A real-world dataset containing human subjects who are motionless, used for detecting inactivity or loss of consciousness.
View Dataset
Edinburgh Simulated Surgical Tools (RGBD)
A synthetic dataset with 2D/3D surgical tools for object recognition and pose estimation.
View Dataset

News & Grants

Apr 2026
Starting my new role at the University of Glasgow!
Sep 2025
📄 Paper nominated for Best Paper Candidate at the Int. Conference on AI in Healthcare, Cambridge.
May 2025
🏆 Won Best Presentation Award at the AI and Dementia International Workshop, Toronto.
Dec 2024
💰 Awarded Moray Endowment Fund (£1.95k), University of Edinburgh.
Feb 2024
💰 Awarded Wellcome Institutional Partnership Award (£5k), University of Edinburgh.

Get In Touch

I am actively seeking new opportunities. Feel free to reach out via the form below or email me directly at kele39@gmail.com.

Visitor Map