CNN Pre-Trained Model Using the Fusion of Features for CBIR Framework

In the era of abundant digital imagery, efficient retrieval of relevant images has become crucial for various applications, including multimedia content management and image analysis. Content-Based Image Retrieval (CBIR) have emerged as a promising solution, leveraging advanced techniques to automatically retrieve images based on their visual content. This research work, proposed a novel CBIR system which exploits the extracted features of fusion from pre-trained model of Convolutional Neural Network (CNN). CNNs have demonstrated remarkable capabilities in learning hierarchical representations of visual features, making them well-suited for image retrieval tasks. By leveraging the rich feature representations learned by a pre-trained CNN, our framework aims to enhance the retrieval accuracy and robustness. We employ techniques for feature fusion to integrate diverse visual cues captured by different layers of the CNN, thus enabling a more comprehensive representation of image content. Furthermore, we proposed methodology which extract the features from two well -known pre-trained CNN like, VGG16 and ResNet50 model. Similarity measurement to effectively match query images with the database. Experimental evaluations conducted on benchmark datasets demonstrate the efficiency of the proposed framework in achieving superior retrieval performance compared to conventional methods. The results underscore the potential of leveraging pre-trained CNN models and feature fusion techniques to advance the state-of-the-art in CBIR systems.