摘要:深度卷积神经网络使用像素级标注，在图像语义分割任务中取得了优异的分割性能.然而，获取像素级标注是一项耗时并且代价高的工作.为了解决这个问题，提出一种基于图像级标注的弱监督图像语义分割方法.该方法致力于使用图像级标注获取有效的伪像素标注来优化分割网络的参数.该方法分为3个步骤：（1）首先，基于分类与分割共享的网络结构，通过空间类别得分（图像二维空间上像素点的类别得分）对网络特征层求导，获取具有类别信息的注意力图；（2）采用逐次擦除法产生显著图，用于补充注意力图中缺失的对象位置信息；（3）融合注意力图与显著图来生成伪像素标注并训练分割网络.在PASCAL VOC 2012分割数据集上的一系列对比实验，证明了该方法的有效性及其优秀的分割性能.



Abstract:Deep convolutional neural networks have achieved excellent performance in image semantic segmentation with strong pixel-level annotations. However, pixel-level annotations are very expensive and time-consuming. To overcome this problem, this study proposes a new weakly supervised image semantic segmentation method with image-level annotations. The proposed method consists of three steps: (1) Based on the sharing network for classification and segmentation task, the class-specific attention map is obtained which is the derivative of the spatial class scores (the class scores of pixels in the two-dimensional image space) with respect to the network feature maps; (2) Saliency map is gotten by successive erasing method, which is used to supplement the object localization information missing by attention maps; (3) Attention map is combined with saliency map to generate pseudo pixel-level annotations and train the segmentation network. A series of comparative experiments demonstrate the effectiveness and better segmentation performance of the proposed method on the challenging PASCAL VOC 2012 image segmentation dataset.



PDF全文下载地址:

http://jos.org.cn/jos/article/pdf/5828