Evaluation of Face Recognition Accuracy in Decoded Images for Machine Vision (Cross-Frame Attention を用いた映像補間モデルの一検討)

Recently, diffusion models are used for video frame interpolation. However, for videos with a lot of movement, the texture collapses when frame interpolation is performed. In this study, we extend the existing research using Cross-Frame Attention to improve the accuracy of frame interpolation.

In Image Information and Television Engineers Annual Convention 2024