In today's world, most of the things we buy, whether we talk about buying apparels, home decor or kitchen utensils, are through the online mode. E-commerce websites play a major role in providing a platform to the vendors for online selling of products. Generally, the process of selling the product on e-commerce websites passes through three steps: Firstly, vendors input the product information including pictures and tag them manually. Secondly, the e-commerce websites perform refinement of tagging using some efficient automated algorithm. Thirdly, the refining of tagging manually and classification of products is done on the basis of their use, popularity, sales, and many more. In this paper, we propose a two-phase process: First, vendor's input images are taken and passed though the deep neural network-based auto-tagging algorithm that tags the input images based on the different features. Next, with the use of dimensionality reduction and clustering algorithms, the tagged images are clustered into different domains. This process helps the vendors and the service provider to automatically tag their product images and provide recommendations on the basis of domain-specific searches without going through the hassle of manual tagging.