A deep learning framework based on modified convolutional neural network for content based image retrieval using elasticsearch integration

With the emergence of cloud-based ecosystem for multimedia content storage, management and processing, there has been tremendous increase in usage of image objects in various real-life applications. For retrieving images in cloud and other storage infrastructures, Content Based Image Retrieval (CBIR) is the process in which an image input is used to search for matching images from the database. This phenomenon is found superior to text-based search due to its modus operandi. However, traditional CBIR methods are based on features comparison and similarity measures. With the emergence of Artificial Intelligence (AI) learning based approach is found to have its benefits to match semantic content as well. Towards this end, in this paper, we proposed a deep learning framework for realizing efficient learning based CBIR system. We proposed a modified CNN model for extracting features from imagery towards better matching experience in image retrieval. We proposed an algorithm known as Learning based Method Content Based Image Retrieval (LbM-CBIR) for realizing the proposed framework. Moreover, the framework has provision to index queries and images into Elasticsearch for making rich experience to users of CBIR. We used ImageNet mini dataset for our experiments. Experimental results revealed that the proposed modified CNN model based CBIR system outperforms existing methods in image retrieval that matches user intent as close as possible. © 2025 Author(s).