Combining the Most Advanced Yet Acceptable (MAYA) Principle with Deep Learning Model to Predict Aesthetic Appreciation of Computers

We propose a novel aesthetic classification framework that combines the Unified Model of Aesthetics (UMA) and the Most Advanced Yet Acceptable (MAYA) principle within an improved YOLOv11s architecture. The model features a PP-LCNet backbone and ShapeIoU loss, enabling fine-grained classification across typical, novel, and MAYA-balanced PC designs. Experiments show strong performance, with a mean Average Precision (mAP@0.5) of 99.42%, precision of 97.9%, and recall of 95.8%. Edge device tests confirm real-time inference capabilities. Moreover, model predictions align closely with human aesthetic judgments. This research contributes to computational aesthetics by offering an intelligent tool for designers, improving product development efficiency, and enabling industries to better align product visuals with user expectations.