Semantic-based image retrieval (SBIR) is a critical challenge at the intersection of natural language processing and computer vision. Traditional retrieval methods primarily depend on metadata annotations or low-level visual feature extraction, often failing to capture user queries' rich contextual and semantic relationships. This study introduces a novel approach that leverages knowledge graphs to enhance SBIR by structuring and representing visual concepts in a more interpretable and relational manner. Specifically, we construct a knowledge graph from the Visual Genome dataset to encode semantic relationships between objects, attributes, and scene compositions. By integrating this knowledge representation into the retrieval process, our approach improves query accuracy, enables more intuitive search mechanisms, and extends the applicability of knowledge graphs in visual information retrieval. Experimental results demonstrate the effectiveness of this method in bridging the semantic gap between textual queries and image content, paving the way for more intelligent and context-aware retrieval systems.