Abstract

Recent advances in saliency detection have utilized deeplearning to obtain high level features to detect salient re-gions in a scene. These advances have demonstrated su-perior results over previous works that utilize hand-craftedlow level features for saliency detection. In this paper, wedemonstrate that hand-crafted features can provide complementary information to enhance performance of saliency detection that utilizes only high level features. Our method utilizes both high level and low level features for saliency detection under a unified deep learning framework. The high level features are extracted using the VGG-net, andthe low level features are compared with other parts of an image to form a low level distance map. The low level distance map is then encoded using a convolutional neural network(CNN) with multiple 1 × 1 convolutional and ReLU layers. We concatenate the encoded low level distance map and the high level features, and connect them to a fully connected neural network classifier to evaluatethe saliency of a query region. Our experiments show that our method can further improve the performance of stateof-the-art deep learning-based saliency detection methods.