PhD Positions on Computer Vision and Deep Learning for Architectural Design

A fair part of our planet is considered as “built environment”. There is a large amount of digital footprint and visual data such as street-view images, maps and interior photos, floor plans, etc. Efficient exploration and analysis of these available data require intelligent tools. Therefore the result is a greater understanding and a superior design of our living environment by architects, designers, and engineers to solve real-world challenges such as sustainability, or health. The Lab’s research agenda focuses on state-of-the-art artificial intelligence methods for automatic recognition and understanding of visual attributes in architectural data: from images of built-forms, to architectural representations and design models. The challenges are the

• Positioning and analysing the state of Computer vision discipline with respect to the demands for efficient and novel computational frameworks in the domain of Architectural design.
• Data efficiency in the context of unsupervised methods for inductive biases such as shapes and image formats in architectural visual forms and data.
• Data collection in the context of the data-hungry supervised-learning models for the visual understanding of the built\design representations.
• Subjectivity in the human-level supervision, e.g., disagreement in the data annotation due to the complex or multifaceted perception of the architectural objects and concepts.
• Communication of the research outcomes to the domain experts (non-computer scientists) particularly in the lack of explainability in the modern machine-learning models.

AiDAPT lab at Architecture offers a Ph.D. position to address the challenges of computer vision for built/design-form recognition and understanding. Your role is to study, analyze and develop deep learning neural network models and the benchmark datasets for visual learning in architectural data. To train and evaluate competent computer vision models, you will work with convolutional neural networks and deep learning programming frameworks such as PyTorch and TensorFlow.

Research Group:
The AiDAPT lab is led by Seyran Khademi (Department of Architecture) and Charalampos Andriotis (Department of Architectural Engineering and Technology) at the Faculty of Architecture and the Built Environment at TU Delft. You will work at the Department of Architecture and in close collaboration with the Computer Vision lab at the faculty of Computer Science.  You’ll be part of the Theory of Architecture and Digital Culture Group that studies the evolving relationship between AI and architecture discipline.