<![CDATA[Content-based image retrieval (CBIR) is an important research area that focuses on emerging techniques for handling large image collections and enabling efficient retrieval. The main challenge of image retrieval is to extract relevant feature vectors for image description. Therefore, visual geometric group 19 (VGG19) with Jaccard is proposed in this research for CBIR. The VGG19 allows to capture of hierarchical features, and it is appropriate for texture and fine detail characteristics. It enables to production of robust and discriminative feature representations because of numerous convolutional layers. The Jaccard is utilized as a similarity measure among feature vectors by comparing the union and intersection of feature sets. It is helpful to manage sparse and higher-dimensional data that provides a fast and accurate image retrieval process. The Caltech 256 and Corel 1K datasets are considered and it is preprocessed by image resizing and normalization. The raw images are resized to ensure dataset similarity and normalized into the range of 0 and 1. The metrics such as recall, f-measure, and average precision are used to calculate the VGG19-Jaccard performance. The VGG19-Jaccard achieves average precision of 99.0 and 99.8% for Caltech 256 and Corel 1K datasets compared to the two-stage CBIR technique.]]>
Copyrights © 2025
