Edge-Aware Convolution Neural Network Based Salient Object Detection

Salient object detection has received great amount of attention in recent years. In this letter, we propose a novel salient object detection algorithm, which combines the global contextual information along with the low-level edge features. First, we train an edge detection stream based on the state-of-the-art holistically-nested edge detection (HED) model and extract hierarchical boundary information from each VGG block. Then, the edge contours are served as the complementary edge-aware information and integrated with the saliency detection stream to depict continuous boundary for salient objects. Finally, we combine pyramid pooling modules with auxiliary side output supervision to form the multi-scale pyramid-based supervision module, providing multi-scale global contextual information for the saliency detection network. Compared with the previous methods, the proposed network contains more explicit edge-aware features and exploit the multi-scale global information more effectively. Experiments demonstrate the effectiveness of the proposed method, which achieves the state-of-the-art performance on five popular benchmarks.

I. Introduction

Visual saliency, which aims to identify the most conspicuous objects or areas in an image, has intrigued great interest in recent years. It has been shown effective owing to various applications in many visual tasks, such as scene classification [1], image captioning [2] and visual tracking [3], [4]. Although visual saliency has achieved state-of-the-art performance through the use of deep convolutional neural networks (CNNs), it still suffers from two major problems in denoising the final prediction against the cluttered background and preserving the boundaries of salient objects.