Deep hash learning for efficient image retrieval


Hashing method is a widely used method for content-based image retrieval. For more complicated semantic similarity of images, supervised hashing methods based on hand-crafted features show its limitations. Convolutional neural network (CNN) has powerful automatic feature learning ability. For this reason, CNN based deep hashing methods outperform previous methods. In this paper, we propose a new deep supervised hashing method for efficient image retrieval. We design a novel deep network with a hash layer as the output layer. An algorithm is proposed to generate optimal target hash code for training. We perform point-wise training for simultaneous feature extracting and hash function learning. Experiments on standard image retrieval benchmarks show that our method outperforms other state-of-the-art methods including unsupervised, supervised and deep hashing methods.

2017 IEEE International Conference on Multimedia Expo Workshops (ICMEW)
Li Song
Li Song
Professor, IEEE Senior Member

Professor, Doctoral Supervisor, the Deputy Director of the Institute of Image Communication and Network Engineering of Shanghai Jiao Tong University, the Double-Appointed Professor of the Institute of Artificial Intelligence and the Collaborative Innovation Center of Future Media Network, the Deputy Secretary-General of the China Video User Experience Alliance and head of the standards group.