An Efficient Hola Filter for Saliency Detection

Abstract

Currently, saliency detection plays an important role in a wide range of applications, such as image segmentation, image recognition, image retrieval, target detection, and so on. These applications require not only the precise saliency localization but also the precise saliency shape. However, most existing approaches did not focus on the precise saliency shape. Therefore, this paper proposes an efficient approach for obtaining the more precise saliency shape. The key contribution of this work is designing a set of Hola filters for more precise localization and sharp edge of detected saliency. Based on a challenging dataset divided into seven categories with different characteristics, experimental results showed that our proposed method outperformed the baselines in almost categories in terms of AUC performance.