Given the exponential growth of global data, research attention has increasingly focused on content-based image retrieval. While deep learning has optimized computer vision tasks, existing image search tools continue to face challenges in processing speed. If these issues are resolved, image search has the potential to fully replace traditional search methods. The drawback of prior methods lies in lengthy feature vectors, resulting in time-consuming searches and inconvenience. To mitigate these issues, we employ feature extraction through convolutional neural networks and subsequently apply binary hashing. Experimental results on the fashion-product-images dataset show that using a convolutional neural network combined with binary hashing in image search achieves a mean average precision (MAP 20) of 0.99 and significantly improves the image retrieval time.