EGNet: Edge Guidance Network for Salient Object Detection (ICCV 2019)

EGNet: Edge Guidance Network for Salient Object Detection

这篇文章发表在ICCV 2019。

FCNs(全卷积网络 fully convotional neual networks)在解决显著性目标检测展现出了自己的优势。但是目前基于全卷积网络的方法有一个弊端——边界轮廓粗糙。本篇文章基于为了解决图像边界的问题,提出了新的框架结构。

本文的三个贡献分别为:

• We propose an EGNet to explicitly model complementary salient object information and salient edge information within the network to preserve the salient object boundaries. At the same time, the salient edge features are also helpful for localization.
• Our model jointly optimizes these two complementary tasks by allowing them to mutually help each other, which significantly improves the predicted saliency maps.
• We compare the proposed methods with 15 state-of-the-art approaches on six widely used datasets. Without bells and whistles, our method achieves the best performance under three evaluation metrics.

本文的大体框架结构如下:
在这里插入图片描述
动机:

基于像素级目标显著性检测方法比基于区域目标显著性方法展示出了其优势。目前也没有方法关注显著边缘检测和显著目标检测的互补性。一个好的目标边缘检测方法在分割和定位中可以帮助显著目标检测任务。基于这个观点,我们提出一个EGNet模型和使用端到端方式的一种单一网络来融合互补的显著边缘信息和显著目标信息。

提出的网络独立于主干网络。主干网络是VGG网络。

实验使用了6个公共数据集。ECSSD,PASCAL-S,DUT-OMRON,SOD, HKU-IS,DUTS.

猜你喜欢

转载自blog.csdn.net/weixin_44790486/article/details/104399165