Feedforward semantic segmentation with zoom-out features
Mohammadreza Mostajabi, Payman Yadollahpour and Gregory Shakhnarovich Toyota Technological Institute at Chicago
Abstract
We introduce a purely feed-forward architecture for semantic segmentation. We map small image elements (superpixels) to rich feature representations extracted from a sequence of nested regions of increasing extent. These regions are obtained by ”zooming out” from the superpixel all the way to scene-level resolution. This approach exploits statistical structure in the image and in the label space without setting up explicit structured prediction mechanisms, and thus avoids complex and expensive inference. Instead superpixels are classified by a feedforward multilayer network. Our architecture achieves new state of the art performance in semantic segmentation, obtaining 64.4% average accuracy on the PASCAL VOC 2012 test set.
中文速览
图像语义分割（semantic segmentation）长期依赖条件随机场等复杂结构化预测模型，推理代价高昂。这篇论文提出了一种纯前向传播（feedforward）的分割方法：把图像先切成超像素（superpixel），然后对每个超像素从局部到全局依次"向外缩放"提取四个尺度的特征——超像素本身、近邻区域、较大区域直到整张图——再把这些"缩放特征"拼接起来，用一个多层神经网络直接做分类，同时引入按类别频率反比加权的非对称损失函数来应对类别不平衡问题。最终该方法在 PASCAL VOC 2012 语义分割测试集上达到 64.4% 的平均准确率，比此前最优方法提升了约 12 个百分点。这项工作的意义在于：它证明了只要特征足够丰富、尺度足够多样，无需显式的结构化推理也能超越依赖复杂图模型的方法，为后续将卷积神经网络直接用于像素级任务提供了重要思路。
原文 arXiv:1412.0774;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1412.0774v1