ProPainter: Improving Propagation and Transformer for Video Inpainting
This work improves video inpainting for applications like video editing and restoration, though it is incremental as it builds on existing propagation and Transformer mechanisms.
The paper tackled limitations in video inpainting by addressing spatial misalignment and computational constraints in propagation and Transformer methods, resulting in ProPainter outperforming prior methods by 1.46 dB in PSNR while maintaining efficiency.
Flow-based propagation and spatiotemporal Transformer are two mainstream mechanisms in video inpainting (VI). Despite the effectiveness of these components, they still suffer from some limitations that affect their performance. Previous propagation-based approaches are performed separately either in the image or feature domain. Global image propagation isolated from learning may cause spatial misalignment due to inaccurate optical flow. Moreover, memory or computational constraints limit the temporal range of feature propagation and video Transformer, preventing exploration of correspondence information from distant frames. To address these issues, we propose an improved framework, called ProPainter, which involves enhanced ProPagation and an efficient Transformer. Specifically, we introduce dual-domain propagation that combines the advantages of image and feature warping, exploiting global correspondences reliably. We also propose a mask-guided sparse video Transformer, which achieves high efficiency by discarding unnecessary and redundant tokens. With these components, ProPainter outperforms prior arts by a large margin of 1.46 dB in PSNR while maintaining appealing efficiency.