Optimizing Footwear Image Classification with Hyperparameter Tuning: Insights from the UTZappos Dataset

In this research, we delve into the realm of footwear image classification utilizing the UTZappos shoe dataset. Employing a suite of convolutional neural network (CNN) architectures, namely ResNet, MobileNet, EfficientNet, and Modified VGG19, we explore their efficacy in accurately categorizing shoe images. Our findings reveal that EfficientNet emerges as the top-performing model, achieving an impressive accuracy rate of 90.02%. This study not only contributes to the field of computer vision and deep learning but also holds practical implications for the fashion industry. By facilitating efficient image sorting and retrieval, our approach can enhance user experience on fashion websites, allowing seamless browsing and discovery of visually similar footwear items, regardless of color variations. Ultimately, this technology has the potential to drive increased engagement and sales by providing consumers with personalized and intuitive shopping experiences.