Salient Object Detection based on CNN Fusion of Two Types of Saliency Models

Salient object detection has attracted significant attention, and recently many heuristic computational models have been developed for object detection. In this paper, we introduced a new approach that is based on convolutional neural network fusion strategy to combine the saliency maps generated by high-dimensional color transform and salient object detection integrating discriminative regional features methods. Our method is based on the observation of salient regions and has a distinctive role in removing background information, which in human’s perspective is not required in the object detection task. The key contribution of this work is a pipeline of CNN based fusion method, which not only performs best but also generates the saliency maps closer to the ground truth. The experimental results show that our method demonstrates the effectiveness of the proposed pipeline while comparing it with other state-of-the-art methods.