Image Completion and Extrapolation with Contextual Cycle Consistency
This addresses the lack of adaptability in existing GAN-based methods for image completion and extrapolation, which is an incremental improvement for computer vision tasks.
The paper tackles the problem of image completion and extrapolation by training networks concurrently to improve adaptability, achieving efficiency in completing large missing regions and showing comparisons with state-of-the-art baselines.
Image Completion refers to the task of filling in the missing regions of an image and Image Extrapolation refers to the task of extending an image at its boundaries while keeping it coherent. Many recent works based on GAN have shown progress in addressing these problem statements but lack adaptability for these two cases, i.e. the neural network trained for the completion of interior masked images does not generalize well for extrapolating over the boundaries and vice-versa. In this paper, we present a technique to train both completion and extrapolation networks concurrently while benefiting each other. We demonstrate our method's efficiency in completing large missing regions and we show the comparisons with the contemporary state of the art baseline.