Diana Davletshina

Research Interests

Diana's research interests include automatic visual scene understanding in the form of segmentation of point clouds and images using computer vision and deep learning, automatic extraction of relationships between objects in a scene, construction of geometric digital twins and keeping them up to date. The application area is the digitisation of existing roads, particularly highways.  

Strategic Themes

• Constructing and maintaining geometric digital twins of highways 

Research Project

My research project focuses on automating the process of constructing and maintaining geometric digital twins of highways using large-scale spatial and visual datasets to reduce related costs and enable further applications such as performance optimisation, failure prediction and future scenario modelling. 

Biography

My background is in software engineering and data science, applied in academia and industry. After obtaining my bachelor's degree in Computer Science from Innopolis University, I worked as a software engineer at the big data department of one of the largest banks. I strengthened my competence in handling massive data there: I designed, developed and deployed frameworks for heterogeneous data processing, synthetic data generation and numerous data-related pipelines' automation. Processed data then was an input for predictive models that exploited inherent patterns and led to solving business problems of the bank. There, I revealed a strong passion for working on data-based inference and artificial intelligence (AI). As such, I obtained a master’s degree in Data Science from the Ludwig Maximilian University of Munich.  

As a result of diving into data science, I accomplished a few projects in the R&D industry in the roles of an applied scientist and team lead. The projects were about computer vision (CV) and deep learning applications in the medical domain. My work facilitated the analysis of 2D and 3D medical images (x-ray, CT, MRI) by automatically highlighting anomalous regions to aid doctors in diagnoses by reducing processing time and helping them avoid missing inconspicuous abnormalities. I also led the development team to deliver a dataset labelling tool.  

Based on my experiences in academia and industry, I am highly motivated to further research and work on the challenges raised by industrial needs in the construction area. My current project aims to digitise existing roads automatically using AI and CV.  

Overall, I have multicultural experience of living, studying and working in my home country, Germany and the UK. My past and current projects are interdisciplinary, and I am eager to bridge the gap between the domains.