Deep Saliency Multi-Task Deep Neural Network Model for Salient Object Detection
Abstract
Multi-Task Deep Saliency .A key problem in salient object detection is how to effectively model the semantic properties of salient objects in a data-driven manner. In this paper, we propose a
Multi-Task Deep Saliency Model based on a fully convolutional neural network with global input (whole raw images) and global output (whole saliency maps). In principle, the proposed saliency model takes a data-driven strategy for encoding the underlying saliency prior information, and then sets up a multi-task learning scheme for exploring the intrinsic correlations between saliency detection and semantic image segmentation.
Through collaborative feature learning from such two correlated tasks, the shared fully convolutional layers produce effective features for object perception.
Moreover, it is capable of capturing the semantic information on salient objects across different levels using the fully convolutional layers, which investigate the feature-sharing properties of salient object detection with a great reduction of feature redundancy.
Finally, we present a graph Laplacian regularized nonlinear regression model for saliency refinement. Experimental results demonstrate the effectiveness of our approach in comparison with the state-of-the-art approaches.
AS AN important and challenging problem in computer vision, salient object detection aims to automatically discover and locate the visually interesting regions that are consistent with human perception.
Conclusion
In this paper, we propose a simple yet effective multitask deep saliency approach for salient object detection based on the fully convolutional neural network with global input (whole raw images) and global output (whole saliency maps). The proposed saliency approach models the intrinsic semantic properties of salient objects in a totally data-driven manner, and performs collaborative feature learning for the two correlated tasks (i.e., saliency detection and semantic image segmentation), which generally leads to the saliency performance improvement in object perception. Moreover, it is capable of accomplishing the feature-sharing task by using a sequence of fully convolutional layers, resulting in a significant reduction of feature redundancy.
