Published Paper Details:

The challenge of content-based image retrieval (CBIR) lies in its reliance on low-level visual features from user query images, making query formulation difficult and often yielding unsatisfactory retrieval results [1] . Previously, image annotation emerged as a promising solution for CBIR, employing automatic assignment of keywords to images for improved retrieval based on user queries. picture annotation essentially mirrors picture class, where low-level features are mapped to excessive-level principles (elegance labels) through supervised mastering algorithms. However, achieving effective feature representations and similarity measures remains critical for CBIR performance. The semantic gap, characterized by the disparity between machine- captured low-level image pixels and human-perceived high-level semantics, poses a significant challenge in this context. Recent advancements in deep learning, particularly Convolutional Neural Networks (CNNs), have demonstrated remarkable success in various computer vision tasks, motivating my pursuit to address the CBIR problem using a dataset of annotated images [6].