arXiv: https://arxiv.org/abs/1912.03992

GitHub: https://github.com/nuneslu/VeIGAN

We proposed a GAN network earlier to deal with disparity data inpainting and object removal in a paper published in the Intelligent Vehicles Symposium 2019 (IV’19) (https://ieeexplore.ieee.org/document/8814157). Now we published on arXiv a more in-depth analysis of our latest network version, which is also open-sourced.

Third column the objects removed using the Contextual Attention network; Last column our results removing the same objects.