Recurrent Attentional Networks for Saliency Detection
This addresses a specific limitation in computer vision for saliency detection, offering an incremental improvement over existing convolutional-deconvolution networks.
The paper tackles the problem of saliency detection for objects of multiple scales by proposing a recurrent attentional convolutional-deconvolution network (RACDNN), which iteratively refines saliency using spatial transformers and recurrent units, and shows it outperforms state-of-the-art methods on challenging datasets.
Convolutional-deconvolution networks can be adopted to perform end-to-end saliency detection. But, they do not work well with objects of multiple scales. To overcome such a limitation, in this work, we propose a recurrent attentional convolutional-deconvolution network (RACDNN). Using spatial transformer and recurrent network units, RACDNN is able to iteratively attend to selected image sub-regions to perform saliency refinement progressively. Besides tackling the scale problem, RACDNN can also learn context-aware features from past iterations to enhance saliency refinement in future iterations. Experiments on several challenging saliency detection datasets validate the effectiveness of RACDNN, and show that RACDNN outperforms state-of-the-art saliency detection methods.