Deep center-based dual-constrained hashing for discriminative face image retrieval
作者：
Highlights：
• A novel center-based deep supervised hashing framework integrating hashing learning and class centers learning for discriminative face image retrieval.
• Cluster intra-class samples into a learnable class center for intra-class variance reduction.
• Enlarge the Hamming distance between pairwise class centers for inter-class separability.
• Regression matrix to enhance binary codes compactness.
• State-of-the-art performance on four large-scale datasets under various code lengths and commonly-used evaluation metrics.
摘要
•A novel center-based deep supervised hashing framework integrating hashing learning and class centers learning for discriminative face image retrieval.•Cluster intra-class samples into a learnable class center for intra-class variance reduction.•Enlarge the Hamming distance between pairwise class centers for inter-class separability.•Regression matrix to enhance binary codes compactness.•State-of-the-art performance on four large-scale datasets under various code lengths and commonly-used evaluation metrics.
论文关键词：Deep supervised hashing,Class centers,Face image retrieval,Convolutional neural networks
论文评审过程：Received 22 June 2020, Revised 23 February 2021, Accepted 30 March 2021, Available online 6 April 2021, Version of Record 16 April 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.107976
