Asymmetric Deep Semantic Quantization for Image Retrieval
share
Due to its fast retrieval and storage efficiency capabilities, hashing has been widely used in nearest neighbor retrieval tasks. By using deep learning based techniques, hashing can outperform non-learning based hashing in many applications. However, there are some limitations to previous learning based hashing methods (e.g., the learned hash codes are not discriminative due to the hashing methods being unable to discover rich semantic information and the training strategy having difficulty optimizing the discrete binary codes). In this paper, we propose a novel learning based hashing method, named Asymmetric Deep Semantic Quantization (ADSQ). ADSQ is implemented using three stream frameworks, which consists of one LabelNet and two ImgNets. The LabelNet leverages three fully-connected layers, which is used to capture rich semantic information between image pairs. For the two ImgNets, they each adopt the same convolutional neural network structure, but with different weights (i.e., asymmetric convolutional neural networks). The two ImgNets are used to generate discriminative compact hash codes. Specifically, the function of the LabelNet is to capture rich semantic information that is used to guide the two ImgNets in minimizing the gap between the real-continuous features and discrete binary codes. By doing this, ADSQ can make full use of the most critical semantic information to guide the feature learning process and consider the consistency of the common semantic space and Hamming space. Results from our experiments demonstrate that ADSQ can generate high discriminative compact hash codes and it outperforms current state-of-the-art methods on three benchmark datasets, CIFAR-10, NUS-WIDE, and ImageNet.
READ FULL TEXT