细粒度图像分类是计算机视觉领域的一大分类任务,其难点在于如何通过类别监督信息自主地找到判别性区域.提出一种新的通道-空间融合注意力模块,基于该模块设计了一种新的Swin Transformer算法SwinT⁃NCSA(a Swin Transformer based on a ...细粒度图像分类是计算机视觉领域的一大分类任务,其难点在于如何通过类别监督信息自主地找到判别性区域.提出一种新的通道-空间融合注意力模块,基于该模块设计了一种新的Swin Transformer算法SwinT⁃NCSA(a Swin Transformer based on a novel channel⁃spatial attention module),分别从通道维和空间维同时提取特征,再将其融入到Swin Transformer模型中以提高其小尺度中多头注意力信息的提取能力.SwinT⁃NCSA算法特别关注了对分类有用的区域,同时忽视对分类无用的背景区域,以此在细粒度图像分类任务中达到较高的分类准确率.在FGVC Aircraft飞机数据集、CUB-200-2011鸟类数据集和Stanford Cars车类数据集3个公共数据集上的实验表明,SwinT⁃NCSA算法可以分别取得93.3%、88.4%和94.7%的准确率,优于同类算法.展开更多
This paper presents a novel algorithm for an extreme form of weak label learning, in which only one of all relevant labels is given for each training sample. Using genetic algorithm, all of the labels in the training ...This paper presents a novel algorithm for an extreme form of weak label learning, in which only one of all relevant labels is given for each training sample. Using genetic algorithm, all of the labels in the training set are optimally divided into several non-overlapping groups to maximize the label distinguishability in every group. Multiple classifiers are trained separately and ensembled for label predictions. Experimental results show significant improvement over previous weak label learning algorithms.展开更多
文摘细粒度图像分类是计算机视觉领域的一大分类任务,其难点在于如何通过类别监督信息自主地找到判别性区域.提出一种新的通道-空间融合注意力模块,基于该模块设计了一种新的Swin Transformer算法SwinT⁃NCSA(a Swin Transformer based on a novel channel⁃spatial attention module),分别从通道维和空间维同时提取特征,再将其融入到Swin Transformer模型中以提高其小尺度中多头注意力信息的提取能力.SwinT⁃NCSA算法特别关注了对分类有用的区域,同时忽视对分类无用的背景区域,以此在细粒度图像分类任务中达到较高的分类准确率.在FGVC Aircraft飞机数据集、CUB-200-2011鸟类数据集和Stanford Cars车类数据集3个公共数据集上的实验表明,SwinT⁃NCSA算法可以分别取得93.3%、88.4%和94.7%的准确率,优于同类算法.
基金Supported by the National Natural Science Foundation of China(61672433)the Fundamental Research Fund for Shenzhen Science and Technology Innovation Committee(201703063000511,201703063000517)+1 种基金the National Cryptography Development Fund(MMJJ20170210)the Science and Technology Project of State Grid Corporation of China(522722180007)
文摘This paper presents a novel algorithm for an extreme form of weak label learning, in which only one of all relevant labels is given for each training sample. Using genetic algorithm, all of the labels in the training set are optimally divided into several non-overlapping groups to maximize the label distinguishability in every group. Multiple classifiers are trained separately and ensembled for label predictions. Experimental results show significant improvement over previous weak label learning algorithms.
