Research

MEng Dissertation (2024)

UCL · The Bartlett School of Architecture

Methods for Applying 2D Architectural Styles to 3D Architectural Models through Machine Learning

Developed a pipeline combining neural style transfer with 3D depth estimation to project 2D architectural styles onto 3D models. Used VGG-19 for style extraction, MiDaS for monocular depth estimation, and custom mesh projection in Python. Awarded publication in the Bartlett Show 2022 Book — sole student selected from the third year.

TensorFlowVGG-19MiDaSMayaviAgisoft MetashapePython
Read full dissertation (PDF) →

Research Interests

  • ◆ Interpretable deep learning and causal reasoning
  • ◆ Affective computing and emotion-aware systems
  • ◆ Human cognition and AI alignment
  • ◆ ML applications in high-frequency / high-stakes environments