arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2012.14123cs.CV
收藏

语义分割的频谱分析及其在特征截断与弱标注中的应用

Spectral Analysis for Semantic Segmentation with Applications on Feature Truncation and Weak Annotation

National Yang Ming Chiao Tung University(国立阳明交通大学)

Li-Wei Chen, Wei-Chen Chiu, Chin-Tien Wu

AI总结:

本文提出频谱分析研究语义分割中下采样分辨率、损失函数与准确性的关系,发现低频分量主导,并应用于特征截断和弱标注以节省成本。

摘要:

众所周知,语义分割神经网络(SSNNs)产生密集的分割图以解决物体边界问题,同时将预测限制在下采样网格上以减轻计算成本。在诸如U-Net等SSNNs的准确性和训练成本之间存在一个显著的平衡。我们提出了一种频谱分析,以研究下采样网格的分辨率、损失函数和SSNNs的准确性之间的相关性。通过分析频域中的网络反向传播过程,我们发现传统的损失函数(交叉熵)和CNN的关键特征主要受分割标签的低频分量影响。我们的发现可以以多种方式应用于SSNNs,包括:(i)确定用于解析分割图的高效低分辨率网格;(ii)通过截断高频解码器特征来修剪网络以节省计算成本;以及(iii)使用分块弱标注以节省标注时间。本文中展示的实验结果与我们对DeepLab V3+和深度聚合网络(DAN)等网络的频谱分析一致。

英文摘要:

It is well known that semantic segmentation neural networks (SSNNs) produce dense segmentation maps to resolve the objects' boundaries while restrict the prediction on down-sampled grids to alleviate the computational cost. A striking balance between the accuracy and the training cost of the SSNNs such as U-Net exists. We propose a spectral analysis to investigate the correlations among the resolution of the down sampled grid, the loss function and the accuracy of the SSNNs. By analyzing the network back-propagation process in frequency domain, we discover that the traditional loss function, cross-entropy, and the key features of CNN are mainly affected by the low-frequency components of segmentation labels. Our discoveries can be applied to SSNNs in several ways including (i) determining an efficient low resolution grid for resolving the segmentation maps (ii) pruning the networks by truncating the high frequency decoder features for saving computation costs, and (iii) using block-wise weak annotation for saving the labeling time. Experimental results shown in this paper agree with our spectral analysis for the networks such as DeepLab V3+ and Deep Aggregation Net (DAN).

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。

↑