MFDANet: Multi-Scale Feature Dual-Stream Aggregation Network for Salient Object Detection

With the development of deep learning, significant improvements and optimizations have been made in salient object detection. However, many detection methods limitations, such as insufficient context information extraction, limited interaction modes for different level features, potential loss due to a single mode. In order solve aforementioned issues, we proposed dual-stream aggregation network based on multi-scale which consists two main modules, namely residual extraction (RCIE) module dense (DDA) module. Firstly, RCIE was designed fully extract by connecting features from receptive fields via connections, where convolutional groups composed asymmetric convolution dilated are used fields. Secondly, DDA aimed enhance relationships between leveraging connections obtain high-quality feature information. Finally, were generate saliency maps. Extensive experiments 5 benchmark datasets show that model performs favorably against 15 state-of-the-art methods.