This product is an open source software. If you would like to access the source code, please visit the below link. (You do not need to click "ADD TO CART" for this item)

<<To ACCESS Source Code: CLICK HERE>>

Authors: Zifan Shi, Na Fan, Dit-Yan Yeung, and Qifeng Chen

Existing vision systems for autonomous driving or robots are sensitive to waterdrops adhered to windows or camera lenses. Most recent waterdrop removal approaches take a single image as input and often fail to recover the missing content behind waterdrops faithfully. Thus, we propose a learning-based model for waterdrop removal with stereo images. To better detect and remove waterdrops from stereo images, we propose a novel row-wise dilated attention module to enlarge attention’s receptive field for effective information propagation between the two stereo images. In addition, we propose an attention consistency loss between the ground-truth disparity map and attention scores to enhance the left-right consistency in stereo images. Because of related datasets’ un-availability, we collect a real-world dataset that contains stereo images with and without waterdrops. Extensive experiments on our dataset suggest that our model outperforms state-of-the-art methods both quantitatively and qualitatively.

Authors:
Department of Computer Science and Engineering Assistance Professor Prof. Qifeng Chen

Key Features:
We have presented a learning-based approach for stereo waterdrop removal, where row-wise dilated attention is proposed to enlarge attention’s receptive field for better left-right information propagation in corrupted stereo images, and attention consistency loss further enhances the consistency in the stereo image pair. To evaluate different methods on stereo real data, we collect a real-world stereo dataset for
waterdrop removal. The experiments have demonstrated that our approach achieves excellent performance on waterdrop
removal with stereo images, as indicated in the user study.
We hope our work can inspire researchers to explore other
image enhancement tasks with stereo images in the future.

Specification:
Please refer to the published paper for specification

Category:
Source Code

Reference:
https://cqf.io/papers/Stereo_Waterdrop_Removal_IROS21.pdf

Contact Us:
ttsamuel@ust.hk