Neighborhood Preserving Hashing for Scalable Video Retrieval

Shuyan Li, Zhixiang Chen, Jiwen Lu, Xiu Li, Jie Zhou. ICCV 2019

[PDF]      
ICCV Video Retrieval Unsupervised Deep Learning LSTM

In this paper, we propose a Neighborhood Preserving Hashing (NPH) method for scalable video retrieval in an unsupervised manner. Unlike most existing deep video hashing methods which indiscriminately compress an entire video into a binary code, we embed the spatial-temporal neighborhood information into the encoding network such that the neighborhood-relevant visual content of a video can be preferentially encoded into a binary code under the guidance of the neighborhood information. Specifically, we propose a neighborhood attention mechanism which focuses on partial useful content of each input frame conditioned on the neighborhood information. We then integrate the neighborhood attention mechanism into an RNN-based reconstruction scheme to encourage the binary codes to capture the spatial-temporal structure in a video which is consistent with that in the neighborhood. As a consequence, the learned hashing functions can map similar videos to similar binary codes. Extensive experiments on three widely-used benchmark datasets validate the effectiveness of our proposed approach.