Robust Representation Learning with Feedback for Single Image Deraining
This work addresses image quality degradation due to rain for computer vision applications, representing an incremental improvement in deraining techniques.
The paper tackles the problem of removing rain streaks from single images by addressing model errors from uncertainty, proposing a method that replaces low-quality features with latent high-quality features using closed-loop feedback. The result shows that the method outperforms recent state-of-the-art methods on benchmark and real datasets.
A deraining network can be interpreted as a conditional generator that aims at removing rain streaks from image. Most existing image deraining methods ignore model errors caused by uncertainty that reduces embedding quality. Unlike existing image deraining methods that embed low-quality features into the model directly, we replace low-quality features by latent high-quality features. The spirit of closed-loop feedback in the automatic control field is borrowed to obtain latent high-quality features. A new method for error detection and feature compensation is proposed to address model errors. Extensive experiments on benchmark datasets as well as specific real datasets demonstrate that the proposed method outperforms recent state-of-the-art methods. Code is available at: \\ https://github.com/LI-Hao-SJTU/DerainRLNet