In the fashion industry, categorizing items based on attributes like color, fabric, size, fit, and comfort is essential for efficient inventory management, trend analysis, and personalized recommendations. Current solutions often rely on manual tagging, which can be both time-consuming and prone to errors. Our project proposes a machine learning model that automates the classification of fashion items into various categories, utilizing publicly available datasets. Dataset Collection and Preprocessing: • Gather and preprocess public datasets containing images of fashion items labeled with attributes like color, fabric, size, fit, and comfort. • Ensure data diversity to account for variations in lighting, angle, and item presentation. Model Development: • Develop a machine learning model capable of multi-label classification, identifying multiple attributes of a fashion item simultaneously. • Experiment with different ML models (ex. Convolutional Neural Networks and Transformers) to achieve optimal accuracy. Model Training and Validation: • Train the model on the collected datasets and validate its performance using appropriate metrics (ex. precision, recall, and F1-score). • Address potential challenges such as imbalanced data, where certain attributes like specific colors or fabrics may be underrepresented.