This work presents Visual-GCN, a hybrid graph-based recommender system tailored for the fashion sector. Visual-GCN combines the collaborative filtering strengths of Graph Convolutional Networks (GCN) with the visual feature analysis derived from product images to improve recommendation accuracy and personalization. Utilizing GCN’s graph-based structure, the model efficiently captures user-item interactions. Concurrently, convolutional neural networks (CNNs) are used to extract stylistic features from fashion images, allowing the system to consider visual item similarities. Experimental results reveal that Visual-GCN surpasses baseline methods by providing trend-aware and personalized recommendations. This study underscores the synergistic advantages of integrating graph-based collaborative filtering with image-based item similarity for fashion recommendations.