Deep feature augmentation for occluded image classification
Deep feature augmentation for occluded image classification
复制标题
DOI：
10.1016/j.patcog.2020.107737
复制
发表时间：
2021-03-01
影响因子：
8
通讯作者：
Wang, Guanghui
中科院分区：
文献类型：
--
作者：
Cen, Feng;Zhao, Xiaoyu;Wang, Guanghui
由于难以获取大量特定于任务的遮挡图像，使用深度卷积神经网络（CNN）对遮挡图像进行分类仍然具有高度挑战性。为了减轻对大规模遮挡图像数据集的依赖，我们提出了一种新的方法，通过使用一组增强的深度特征向量（DFV）来微调预训练模型，以提高遮挡图像的分类准确性。增广的DFV集合由原始DFV和伪DFV组成。伪DFV是通过将从一小组干净和被遮挡的图像对中提取的差向量（DV）随机添加到真实的DFV来生成的。在微调中，对DFV数据流进行反向传播以更新网络参数。在各种数据集和网络结构上的实验表明，深度特征增强显著提高了遮挡图像的分类精度，而对干净图像的性能没有明显影响。具体来说，在ILSVRC 2012数据集与合成闭塞图像，所提出的方法实现了11.21%和9.14%的平均提高分类精度的ResNet50网络上的闭塞排他性和闭塞包容性的训练集，分别微调。(C)2020爱思唯尔有限公司保留所有权利。
Due to the difficulty in acquiring massive task-specific occluded images, the classification of occluded images with deep convolutional neural networks (CNNs) remains highly challenging. To alleviate the dependency on large-scale occluded image datasets, we propose a novel approach to improve the classification accuracy of occluded images by fine-tuning the pre-trained models with a set of augmented deep feature vectors (DFVs). The set of augmented DFVs is composed of original DFVs and pseudo-DFVs. The pseudo-DFVs are generated by randomly adding difference vectors (DVs), extracted from a small set of clean and occluded image pairs, to the real DFVs. In the fine-tuning, the back-propagation is conducted on the DFV data flow to update the network parameters. The experiments on various datasets and network structures show that the deep feature augmentation significantly improves the classification accuracy of occluded images without a noticeable influence on the performance of clean images. Specifically, on the ILSVRC2012 dataset with synthetic occluded images, the proposed approach achieves 11.21% and 9.14% average increases in classification accuracy for the ResNet50 networks fine-tuned on the occlusion-exclusive and occlusion-inclusive training sets, respectively. (C) 2020 Elsevier Ltd. All rights reserved.