Application of Computer Vision Approach for Automation and Classification of Fashion Store

The fashion industry has long been interested in utilizing computer vision techniques to automate tasks such as recognizing different types of clothing items in images. This study proposed a novel convolutional neural network (CNN) architecture for the classification of fashion outfits into three categories: shoes, sunglasses, and trousers. The proposed CNN architecture is based on state-of-the-art deep learning techniques and is trained on a fairly large-scale dataset of fashion images. The effectiveness of the proposed CNN architecture is evaluated through extensive experiments and analysis. The result demonstrates that the proposed CNN architecture achieves a high accuracy rate of 0.99 on each diagonal value of the confusion matrix. This indicates that the proposed CNN is capable of accurately classifying each item type with high accuracy. Additionally, the study investigates the impact of various hyperparameters on the performance of the proposed CNN architecture and found that the model which uses a 7x7 filter size and 64 filter number yields higher accuracy compared to other filter combinations. The study demonstrates the potential of CNN in automating fashion item recognition, which can lead to improved efficiency and accuracy in the fashion industry. It can also form the basis for developing more advanced computer vision systems for the fashion industry.