Image-Based Prediction of Perceived Spatial Functionality (PSF) in Design Studios

This repository contains the data associated with the research: "Image-Based Prediction of Perceived Spatial Functionality in Design Studios Using Deep Learning" by Natnael Wubishet Bahiru. The research develops a computational framework to quantify the subjective experience of architectural interiors. It evaluates whether a specialized deep learning regressor (based on the DINOv2 Vision Transformer) can approximate human ratings of "Perceived Spatial Functionality" (PSF) in design studio environments, specifically identifying a "Perception Gap" between generalist Vision-Language Models (VLMs) and human architectural intuition.