arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1601.07576cs.CV
收藏

局部监督深度混合场景识别模型

Locally-Supervised Deep Hybrid Model for Scene Recognition

University of Chinese Academy of Sciences(中国科学院大学);Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences(中国科学院深圳先进技术研究院);ETH Zurich(苏黎世联邦理工学院);The Chinese University of Hong Kong(香港中文大学)

Sheng Guo, Weilin Huang, Limin Wang, Yu Qiao

摘要:

卷积神经网络(CNN)最近在各种图像分类和理解任务中取得了显著的成功。在CNN的顶部全连接层处获得的深度特征(FC特征)表现出丰富的全局语义信息,并且在图像分类中非常有效。另一方面,CNN的中间层中的卷积特征也包含有意义的局部信息,但没有完全用于图像表示。在本文中,我们提出了一种新的局部监督深度混合模型(LS-DHM),它有效地增强和探索了用于场景识别的卷积特征。首先,我们注意到卷积特征捕获了场景图像的局部对象和精细结构,这为区分模糊场景提供了重要线索,而这些特征在高度压缩的FC表示中被显著消除。其次,我们提出了一个新的局部卷积监督(LCS)层,通过直接将标签信息传播到卷积层来增强图像的局部结构。第三,我们提出了一个有效的Fisher卷积向量(FCV),成功地拯救了无序的中层语义信息(如对象和纹理)的场景图像。FCV将大型卷积映射编码为固定长度的中级表示,并被证明是对高级FC特征的有力补充。最后,FCV和FC-功能的LSDHM表示,这在我们的实验中取得了优异的性能。在MIT Indoor 67和SUN 397数据集上分别获得了83.75%和67.56%的准确率,大大提高了最先进的水平。

英文摘要:

Convolutional neural networks (CNN) have recently achieved remarkable successes in various image classification and understanding tasks. The deep features obtained at the top fully-connected layer of the CNN (FC-features) exhibit rich global semantic information and are extremely effective in image classification. On the other hand, the convolutional features in the middle layers of the CNN also contain meaningful local information, but are not fully explored for image representation. In this paper, we propose a novel Locally-Supervised Deep Hybrid Model (LS-DHM) that effectively enhances and explores the convolutional features for scene recognition. Firstly, we notice that the convolutional features capture local objects and fine structures of scene images, which yield important cues for discriminating ambiguous scenes, whereas these features are significantly eliminated in the highly-compressed FC representation. Secondly, we propose a new Local Convolutional Supervision (LCS) layer to enhance the local structure of the image by directly propagating the label information to the convolutional layers. Thirdly, we propose an efficient Fisher Convolutional Vector (FCV) that successfully rescues the orderless mid-level semantic information (e.g. objects and textures) of scene image. The FCV encodes the large-sized convolutional maps into a fixed-length mid-level representation, and is demonstrated to be strongly complementary to the high-level FC-features. Finally, both the FCV and FC-features are collaboratively employed in the LSDHM representation, which achieves outstanding performance in our experiments. It obtains 83.75% and 67.56% accuracies respectively on the heavily benchmarked MIT Indoor67 and SUN397 datasets, advancing the stat-of-the-art substantially.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑