Deep Learning Researcher

Dr Benedikt
Kolbeinsson

My research focuses on advancing deep learning methods and developing novel approaches to solve challenging problems. I specialise in computer vision, particularly for fully autonomous drones, LLMs, RAG, diffusion models and knowledge distillation.

Stylised portrait of Benedikt Kolbeinsson
01. Research

Featured Research

Exploring the intersection of computer vision, autonomous systems and artificial intelligence through innovative methods and real-world applications.

Research Overview

My research spans multiple domains within artificial intelligence, with particular emphasis on practical applications that can solve real-world challenges. I focus on developing novel, robust systems that can operate in complex, dynamic environments.

Multi-Class Segmentation from Aerial Views using Recursive Noise Diffusion

WACV 2024 Computer Vision Diffusion Models

Novel approach to multi-class segmentation for aerial imagery using recursive noise diffusion techniques, advancing the state-of-the-art in autonomous drone vision systems.

DDOS: The Drone Depth and Obstacle Segmentation Dataset

CVPR 2024 Dataset Autonomous Drones

Comprehensive dataset for drone depth estimation and obstacle segmentation, providing the research community with essential resources for advancing autonomous aerial navigation.

UCorr: Wire Detection and Depth Estimation for Autonomous Drones

ROBOVIS 2024 Safety Detection

Advanced wire detection system for autonomous drones, addressing critical safety challenges in aerial navigation through improved depth estimation and obstacle detection.

More publications available on Google Scholar

02. Education

Academic Background

A foundation in engineering, refined through advanced study and research.

PhD • Imperial College London

Deep Learning for Autonomous Drone Vision

Doctor of Philosophy • 2025

Research in AI and machine learning, focusing on computer vision and autonomous systems. Developed novel deep learning architectures and datasets, with publications in top-tier conferences.

MEng • Imperial College London

Electrical and Electronic Engineering with Management

Master of Engineering • 2019

Thesis on GANs for domain transfer, specialised in machine learning and computer vision.

Stúdentspróf • Verzlunarskóli Íslands

Stúdentspróf - Náttúrufærðibraut, eðlisfræðisvið

Verzlunarskóli Íslands • 2015

Physics track with foundation studies in mathematics and sciences, establishing early academic excellence in analytical thinking and problem-solving.

03. Experience

Professional Journey

Bridging the gap between cutting-edge research and real-world applications through entrepreneurship and collaborative innovation.

Co-founder & CTO • Regava

AI-Driven Regulatory Compliance

June 2024 - Present

Leading research and development of LLM-powered solutions for regulatory compliance in financial services. Developing AI-driven platform that intelligently navigates complex financial regulations through proactive horizon scanning and real-time compliance analysis.

Research Scientist • Deepair

AI Solutions for Travel Industry

May 2018 - June 2022 (Part-time)

Developed and deployed reinforcement learning systems for automated pricing in the travel industry. Presented research on fairness in reinforcement learning at NeurIPS 2020 workshop, addressing real-world challenges in dynamic pricing systems.

Data Scientist • Icelandic COVID-19 Vaccine Taskforce

Public Health Impact

April 2021 - June 2021

Led data science initiatives that increased daily vaccination rates from 5k to 17k, implemented live monitoring systems, and optimised patient recruitment through predictive analytics during Iceland's national vaccination programme.

Current Focus

Balancing entrepreneurial ventures with cutting-edge research, working to translate advanced AI research into practical solutions that can benefit society and industry.

04. Recognition

Awards & Achievements

Recognition for excellence in research, innovation and academic achievement across multiple institutions and organisations.

Chartered Engineer

Ministry of Higher Education, Science and Innovation • 2020

Winner of the Huawei Autonomous Vehicles Challenge

Huawei Technologies • 2019

Best One Minute Thesis

Imperial College London

Best Third Year Project

Imperial College London • 2018

Outstanding Achievement in Mathematics

Íslenska Stærðfræðafélagið (The Icelandic Mathematical Society)

Associateship of the City and Guilds of London Institute (ACGI)

City and Guilds of London Institute

05. Connect

Get in Touch

Feel free to connect through these platforms.