Generative image inpainting with submanifold alignment
A Li, J Qi, R Zhang, X Ma, K Ramamohanarao
Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence | International Joint Conference on Artificial Intelligence | Published : 2019
Image inpainting aims at restoring missing regions of corrupted images, which has many applications such as image restoration and object removal. However, current GAN-based generative inpainting models do not explicitly exploit the structural or textural consistency between restored contents and their surrounding contexts. To address this limitation, we propose to enforce the alignment (or closeness) between the local data submanifolds (or subspaces) around restored images and those around the original (uncorrupted) images during the learning process of GAN-based inpainting models. We exploit Local Intrinsic Dimensionality (LID) to measure, in deep feature space, the alignment between data s..View full abstract
Related Projects (1)
Awarded by Australian Research Council
The work is partially supported by the ARC grant DP170103174 and by the China Scholarship Council.