Paper 07 - Localization of Deep Video Inpainting Based on Spatiotemporal Convolution and Refinement Network