Ransp: Ranking Attention Network For Saliency Prediction On Omnidirectional Images
Ransp: Ranking Attention Network For Saliency Prediction On Omnidirectional Images
复制标题
DOI：
10.1109/icme46284.2020.9102867
复制
发表时间：
2020-07
期刊：
影响因子：
--
通讯作者：
Dandan Zhu;Yongqing Chen;Tian Han;Defang Zhao;Yucheng Zhu;Qiangqiang Zhou;Guangtao Zhai;Xiaokang Yang
中科院分区：
文献类型：
--
作者：
Dandan Zhu;Yongqing Chen;Tian Han;Defang Zhao;Yucheng Zhu;Qiangqiang Zhou;Guangtao Zhai;Xiaokang Yang
各种基于卷积神经网络(CNN)的方法已经显示出能够提高全向图像(ODI)上显著预测的性能。然而，这些方法受到次优精度的限制，因为并不是所有由CNN模型提取的特征都对最终的细粒度显著预测有用。特征是冗余的，并且对最终的细粒度显著预测有负面影响。为了解决这一问题，我们提出了一种新的基于注意力网络的头部注视显著预测方法。具体地说，部分引导注意(PA)模块和通道特征(CF)提取模块被集成在一个统一的框架中，并以端到端的方式进行训练，以进行细粒度的显著预测。为了更好地利用通道特征图，我们进一步提出了一种新的排序注意模块(RAM)，它根据评分自动对这些图进行排序和选择，以进行细粒度的显著预测。大量的实验表明，该方法对ODI的显著预测是有效的。
Various convolutional neural network (CNN)-based methods have shown the ability to boost the performance of saliency prediction on omnidirectional images (ODIs). However, these methods are limited by sub-optimal accuracy, because not all the features extracted by the CNN model are not useful for the final fine-grained saliency prediction. Features are redundant and have negative impact on the final fine-grained saliency prediction. To tackle this problem, we propose a novel Ranking Attention Network for saliency prediction (RANSP) of head fixations on ODIs. Specifically, the part-guided attention (PA) module and channel-wise feature (CF) extraction module are integrated in a unified framework and are trained in an end-to-end manner for fine-grained saliency prediction. To better utilize the channel-wise feature map, we further propose a new Ranking Attention Module (RAM), which automatically ranks and selects these maps based on scores for fine-grained saliency prediction. Extensive experiments are conducted to show the effectiveness of our method for saliency prediction of ODIs.