How to search images people are interested in quickly and accurately among large-scale image and video data becomes a challenge. The proposed deep learning combined with the hash function retrieval method can not only learn the advanced features of the image to eliminate the need to manually design the feature extractor, but also reduce dimension to improve the retrieval effect. Firstly, improve GoogLeNet neural network to extract the high-level features of pictures. Secondly, add the hash function hidden layer at different levels of the network to make up for the lack of detailed information of high-level features, and fuse the different features of the image to generate hash binary encoding with certain weights. Finally, the images are sorted by similarity according to the Hamming distance and the method would return similar images. The experimental results show that the proposed method has a significant improvement in accuracy and efficiency on the CIFAR-10 and NUS-WIDE datasets.
CITATION STYLE
Zhang, J., Li, Y., & Zeng, Z. (2020). Improved Image Retrieval Algorithm of GoogLeNet Neural Network. In Advances in Intelligent Systems and Computing (Vol. 1063, pp. 25–34). Springer. https://doi.org/10.1007/978-3-030-31967-0_3
Mendeley helps you to discover research relevant for your work.