引用本文: | 刘萌,周迪,田传发,等.上下文感知的深度弱监督图像哈希表示学习方法[J].国防科技大学学报,2022,44(3):85-92.[点击复制] |
LIU Meng,ZHOU Di,TIAN Chuanfa,et al.Context-aware deep weakly supervised image hashing learning method[J].Journal of National University of Defense Technology,2022,44(3):85-92[点击复制] |
|
|
|
本文已被:浏览 3553次 下载 2835次 |
上下文感知的深度弱监督图像哈希表示学习方法 |
刘萌,周迪,田传发,齐孟津,聂秀山 |
(山东建筑大学 计算机科学与技术学院, 山东 济南 250101)
|
摘要: |
针对现有深度监督图像哈希表示学习方法依赖于图像的类别信息,难以在现实中被广泛应用问题,利用与图像相关的标签信息作为监督信息,提出上下文感知的深度弱监督图像哈希表示学习方法。该方法一方面通过自适应捕获图像区域特征的相关上下文来增强它们的表示能力,另一方面通过引入判别损失来提高学习到的哈希码表示的判别性。在现有两个公开数据集上的大量实验结果证明了该方法的有效性。 |
关键词: 图像哈希 弱监督学习 图像检索 区域上下文建模 判别损失 |
DOI:10.11887/j.cn.202203011 |
投稿日期:2021-06-07 |
基金项目:国家自然科学基金资助项目(62006142) |
|
Context-aware deep weakly supervised image hashing learning method |
LIU Meng, ZHOU Di, TIAN Chuanfa, QI Mengjin, NIE Xiushan |
(School of Computer Science and Technology, Shandong Jianzhu University, Jinan 250101, China)
|
Abstract: |
Existing deep supervised image hashing approaches rely on substantial labeled image data, which is very difficult to be widely applied in reality. By utilizing tags associated with images as the supervision information, a context-aware deep weakly supervised image hashing method was proposed. The method enhanced the image region representations by adaptively capturing the relevant context information of image region features, and raised the discrimination of the learnt hash codes by introducing a discrimination loss. Extensive experiments on two public datasets show the effectiveness of the method. |
Keywords: image hashing weakly supervised learning image retrieval region context modeling discrimination loss |
|
|
|
|
|