Dual Attention-in-Attention Model for Joint Rain Streak and Raindrop Removal

Rain streaks and rain drops are two natural phenomena, which degrade image capture in different ways. Currently, most existing deep deraining networks take them as distinct problems individually address one, thus cannot deal adequately with both simultaneously. To this, we propose a Dual Attention-in-Attention Model (DAiAM) includes DAMs for removing raindrops. Inside the DAM, there attentive maps - each of attends to heavy light rainy regions, respectively, guide process differently applicable regions. In addition, further refine result, Differential-driven (D-DAiAM) is proposed "heavy-to-light" scheme remove via addressing unsatisfying Extensive experiments on one public raindrop dataset, streak our synthesized joint (JRSRD) dataset have demonstrated that method not only capable raindrops simultaneously, but also achieves state-of-the-art performance tasks.