Overview of Image Inpainting Techniques: A Survey
Riya Shah, Anjali Gautam, Satish Kumar Singh · 2022 IEEE Region 10 Symposium (TENSYMP) · 2022
Images that are corrupted during transmission or improper storage, can be restored via image inpainting by filling the missing/damaged region of images such that the observer can-not perceive the inpainted region. This paper provides an overview of the traditional methods and deep learning methods which have been used for inpainting task. The traditional methods could accurately fill the missing regions when the hole size is small but fails to inpaint large sized holes and also they cannot hallucinate novel contents. With the availability of huge computation power systems, the recent advancements in deep learning methods have shown exceptional results in image inpainting tasks. However, there is still room for improvement in this task in terms of the mask size of arbitrary shape and at arbitrary locations, reducing computational resources, reducing training time, generating high quality results etc.