Batch covariance neural network for image recognition
Batch covariance neural network for image recognition
复制标题
用于图像识别的批量协方差神经网络
DOI：
10.1016/j.imavis.2022.104446
复制
发表时间：
2022-04
影响因子：
4.7
通讯作者：
Xiaotian Lin
中科院分区：
文献类型：
--
作者：
Tianyou Zheng;Qiang Wang;Yue Shen;Xiang Ma;Xiaotian Lin
最近的工作表明，如果数据集建立得很好，卷积神经网络(CNN)可以达到最先进的水平。然而，现有的卷积层受到各种数据集的影响，不可避免地存在局部异常特征问题，即光照强度和特征相互作用。本文用批处理协方差层(BCL)代替卷积层来定位不受问题影响的类别相关区域。BCL被看作是一种三维协方差运算，它计算所有通道的核大小与特征映射之间的相关性。描述了BCL的前向传播、后向传播、梯度更新和测试过程。BCL与卷积层的比较表明，BCL能够减少光照强度和特征交互对识别和生成任务的影响。复杂度分析表明，BCL算法可以在略微增加时间消耗的情况下提高精度。此外，批处理协方差神经网络(BCovNN)是在CNN的基础上扩展的，用批处理协方差神经网络代替卷积层。烧蚀实验验证了BCovNN的改进是由BCL单独提供的。BCovNN在几个流行的数据集(即MNIST、STL-10、CIFAR-10和ImageNet)上进行了评估，用于图像识别和Pascal VOC(2007和2012)数据集上的对象定位。实验结果表明，BCovNN比相应的CNN有明显的改进。
Recent work has shown that convolutional neural networks (CNN) can achieve state of the art if the datasets are well built. However, the existing convolutional layer is affected by various datasets with the inevitable problems of local abnormal features, i.e., illumination intensity and feature interaction. This paper replaces the convolutional layer with a batch covariance layer (BCL) to locate the category-related region unaffected by the problems. The BCL is regarded as a 3D covariance operation, which calculates the correlation between the kernels and feature maps in kernel size of all channels. Forward propagation, backward propagation, gradient updating, and testing procedure of the BCL are described. The comparison between BCL and convolutional layer shows the ability of BCL to reduce the influence of illumination intensity and feature interaction for discriminating and generating tasks. Complexity analysis shows that BCL can improve the accuracy with a thimbleful time consumption increase. Besides, the batch covariance neural network (BCovNN) is extended from the CNN by replacing the convolutional layer with BCL. Ablation experiment verifies the improvement of BCovNN is provided by BCL separately. BCovNN is evaluated on several popular datasets (i.e., MNIST, STL-10, CIFAR-10, and ImageNet) for image recognition and PASCAL VOC (2007 and 2012) datasets for object localization. Experimental results reveal that BCovNN achieves significant improvements over the corresponding CNN.
登录
查看更多内容
DOI：
10.1109/icccnt56998.2023.10306417
发表时间：
2022-02
期刊：
2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT)
影响因子：
--
作者：
Gilad Cohen;Raja Giryes
通讯作者：
Gilad Cohen;Raja Giryes
DOI：
--
发表时间：
2017-02
期刊：
ArXiv
影响因子：
--
作者：
M. Babaee;D. Dinh;G. Rigoll
通讯作者：
M. Babaee;D. Dinh;G. Rigoll
DOI：
10.1109/cvpr.2016.514
发表时间：
2016-04
期刊：
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
影响因子：
--
作者：
Lingxi Xie;Jingdong Wang;Zhen Wei;Meng Wang;Qi Tian
通讯作者：
Lingxi Xie;Jingdong Wang;Zhen Wei;Meng Wang;Qi Tian
影响因子：
19.5
作者：
Russakovsky, Olga;Deng, Jia;Fei-Fei, Li
通讯作者：
Fei-Fei, Li
DOI：
10.1109/tpami.2016.2567384
发表时间：
2017-04-01
影响因子：
23.6
作者：
Dosovitskiy, Alexey;Springenberg, Jost Tobias;Brox, Thomas
通讯作者：
Brox, Thomas