Visual image recommendation uses artificial intelligence to suggest similar products based on their appearance. In e-commerce, this allows consumers to easily find items matching their aesthetic preferences, thus improving user experience and increasing conversion rates. Companies can personalize their offerings and retain customers by catering precisely to their visual tastes. This approach can be applied to various fields such as e-commerce, social networks, art and design, and tourism. Clustering, is an unsupervised learning technique used to group similar objects into clusters. In the context of recommendation, clustering can group users or objects (like products, movies, articles, etc.) based on their characteristics and behavior. In this article, we will conduct a comparative study of clustering techniques for visual recommendation and examine their effects on this system. The k-means, KMedoids, MiniBatch k-means models that we compared showed that the K-means model outperformed others with a silhouette score of 0.0921, surpassing all methods on our Fashion Product Image dataset [1].