Name
Abstract | ||
|---|---|---|
In this paper, we present an efficient deep supervised hashing method to learn robust hash codes for content-based image retrieval on large-scale datasets. Deep hashing methods have achieved some good results in image retrieval by training the network with classification loss and constructing hash functions as a latent layer. However, the classification loss does not impose a sufficient constraint on the network to make sure that similar images can be encoded to similar binary codes. As a supplement to classification loss, a new loss is delicately designed in our method. After trained with the joint objective functions, the network can generate more discriminative hash codes, which will increase the performance of retrieval. Our method outperforms the state-of-the-art methods by an obvious margin on three datasets CIFAR-10, CIFAR-100 and MNIST. Especially, the improvement is more impressive when the code length is short and the category number is large. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1007/978-3-030-00776-8_24 | ADVANCES IN MULTIMEDIA INFORMATION PROCESSING, PT I |
Keywords | Field | DocType |
Image retrieval,Deep hashing,Discriminative quantization hashing | MNIST database,Pattern recognition,Computer science,Binary code,Image retrieval,Hash function,Artificial intelligence,Quantization (signal processing),Discriminative model | Conference |
Volume | ISSN | Citations |
11164 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 0 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jingbo Fan | 1 | 0 | 0.68 |
| Chuanchuan Chen | 2 | 0 | 0.68 |
| Zhu Yuesheng | 3 | 112 | 39.21 |