iSukat: Capture, measure, and know your perfect shoe size using image recognition

ABSTRACT: Improper shoe sizing remains a widespread issue in online footwear shopping, with studies indicating that approximately 63-72% of consumers wear incorrectly sized shoes. This often leads to discomfort, dissatisfaction, and potential long-term foot health problems. In response, iSUKAT presents a web-based application that harnesses advanced image recognition and machine learning technologies to deliver accurate foot measurements and improve shoe selection for users. By utilizing Roboflow 3.0 with COCO-seg instance segmentation, the system processes foot images to extract precise dimensions, identify foot types such as Egyptian, Roman, and Germanic, and recognize shoe brands through visual pattern detection. The study addresses key challenges in digital shoe fitting, aiming to minimize sizing errors and enhance user satisfaction through automated, intelligent recommendations. The system supports seamless performance across devices, offering a user-friendly interface and real-time processing to ensure smooth operation. Evaluation using machine learning metrics such as mean average precision, segmentation accuracy, and inference time confirmed the model’s efficiency. Despite some concerns-where only 11.6% of users strongly agreed with the measurement accuracy while 40.6% raised issues----ISUKAT will demonstrated consistent performance in detection and recommendation.

Filipiniana