Deep hash Based on Asymmetric Learning and Center Contrast

Benefiting from the storage and retrieval efficiency, learning to hash has been widely used in large-scale multimedia retrieval. Deep learning to hash has attracted the attention of many scholars because it can better extract image features while achieving end-to-end encoding. In recent years, many scholars have focused their research on hash-centered methods that can learn to use global features. Based on the hash center method, this paper uses contrast loss and asymmetric learning methods to enable convolutional neural networks to further learn global features. The experimental part shows that the method proposed in this paper can achieve good results.