The widespread usage of mobile devices has substantially increased the volume of image data gathered by users. This has led to a rising emphasis on image retrieval, which generally depends on either handcrafted feature extraction or deep learning-based algorithms. This research provides a hybrid image retrieval technology that blends handcrafted features, based on color and texture, with deep features acquired from deep neural networks. The results demonstrate that integrating handcrafted features with deep features, specifically those extracted using EfficientNet-B7 model and evaluated with cosine distance measure, significantly enhances retrieval performance on corel-1 k, achieving an impressive precision of 98.85%. By comparing the outcomes through various methodologies, we efficiently analyzed the outcomes, which produced better results. The model's future improvement could involve utilizing nearest neighbor search techniques, clustering strategies, and deep neural network refinement.