About

I am a Postdoctoral researcher in Deep Learning and Computer Vision at the University of Oxford, working with Prof Ben Sheldon. My research focuses on designing innovative deep learning architectures that advance feature representation learning — particularly for extracting salient features in complex, densely populated scenes.

Additionally, I investigate multimodal deep learning and data fusion techniques, exploring how combining diverse modalities — such as RGB, hyperspectral, and thermal imagery — can improve model generalisation across domains. By advancing deep learning architectures, my work drives methodological innovation, validated through remote sensing.

My current research portfolio includes:

Introducing a novel semantic segmentation architecture featuring custom convolutional layers that enhance feature extraction in complex visual domains, advancing segmentation robustness beyond standard methods.
Developing hybrid CNN-Transformer architectures that redefine feature attention mechanisms, optimising the synergy between convolutional feature locality and Transformer-based global context modelling.
Designing multimodal mid-level data fusion architectures for industrial monitoring using multi-temporal and multi-spectral remotely sensed data.
Creating robust segmentation models for UAV-based waterbody semantic segmentation, resilient to occlusion and shadowing.
Enhancing deep learning robustness to seasonal domain shifts through invariant feature representation.

To advance deep learning model development, I curate high-quality datasets for various computer vision tasks across diverse environments.

Background:

Prior to this, I completed an EPSRC-funded PhD at the University of Sussex under Prof. Andy Philippides and Prof. Novi Quadrianto. My research focused on salient feature representation in domain adaptation and semantic segmentation, leading to the development of new models for remote sensing applications.

As part of the Predictive Analytics Lab (PAL), I contributed to a British Academy-funded project on Satellite/Aerial Image Scene Segmentation. I designed novel deep learning models for land-use classification from imagery, integrating multiple data types into a web-based mapping and visualisation tool for peri-urban agriculture in Ghaziabad, India. Please see the demo below. demo During my research internship at Satellite Applications Catapult (supervised by Dr Cristian Rossi), I applied physics-aware AI and remote sensing data to detect and classify cement plants in China by leveraging physical properties such as ambient temperature and soil moisture.

Industrial Experience:

Prior completing my PhD, I worked as a Control and Automation Design Engineer in the oil and gas industry, gaining experience across the full software and hardware development lifecycle — including systems sensing and control, communication protocols, functional safety, and project planning.

Open to Collaboration:

I am always eager to explore new collaborations — Feel free to get in touch if you would like to connect!

Dr Georgios Voulgaris

Recent News