<![CDATA[The coffee shop industry is increasingly competitive, requiring business owners to adopt data-driven strategies rather than rely solely on intuition, as traditional approaches such as surveys and manual observation are often subjective, time-consuming, and lack scalability. This study aims to design, implement, and evaluate an end-to-end intelligent system based on computer vision to automatically and non-intrusively analyze customer demographics (age and gender) and visit duration (dwell time). The proposed framework emphasizes both technical accuracy and privacy-by-design principles, where facial data is processed in real time without storage, and only anonymized metadata is utilized for business analysis. Using a simulated 60-minute test video containing 50 virtual customers with balanced gender, varied age groups, and predetermined visit durations, the system was evaluated and demonstrated strong performance, achieving 96% accuracy in gender classification, 89% in age group classification, and a Mean Absolute Error (MAE) of less than 45 seconds in dwell time measurement. The findings confirm that the ethical application of computer vision can provide valuable business insights, including the identification of demographic-based peak hours, the recognition of product preferences, and the optimization of spatial layouts, ultimately enabling coffee shops and SMEs to enhance competitiveness and profitability through data-driven decision-making.]]>
Copyrights © 2025
