Keywords
feedback, lighting, composition

Technologies
CLIP, Figma, Kotlin

Urban Aesthetics is an Android application developed to support a psychology study examining whether AI-generated photographic feedback can improve users' photographic outcomes. The application enables users to submit a photograph and receive structured, interpretable feedback across five visual dimensions: lighting/exposure, contrast, framing/perspective, visual focus, and visual complexity. Scores are generated using a zero-shot CLIP ViT-B/16 pipeline, in which per-feature deltas derived from cosine similarity comparisons between the image embedding and positive and negative prompt sets are mapped to a calibrated 0–10 scale. A large language model then translates the numeric evidence into plain-English summaries and actionable suggestions for each feature. The backend is implemented as a FastAPI service; the front end is a Jetpack Compose Android application. The system was designed to make photographic feedback decomposable, transparent, and actionable rather than reducible to a single opaque score.