Enhancing Deformable Convolution based Video Frame Interpolation with Coarse-to-fine 3D CNN

TL;DR

This paper introduces a deformable convolution-based video frame interpolation method utilizing a coarse-to-fine 3D CNN to improve multi-flow prediction, resulting in superior interpolation quality compared to existing methods.

Contribution

The paper proposes a novel coarse-to-fine 3D CNN framework for deformable convolution-based video frame interpolation, enhancing multi-flow estimation and interpolation accuracy.

Findings

Outperforms 12 state-of-the-art VFI methods

Achieves PSNR gains up to 0.19dB

Demonstrates superior interpolation performance

Abstract

This paper presents a new deformable convolution-based video frame interpolation (VFI) method, using a coarse to fine 3D CNN to enhance the multi-flow prediction. This model first extracts spatio-temporal features at multiple scales using a 3D CNN, and estimates multi-flows using these features in a coarse-to-fine manner. The estimated multi-flows are then used to warp the original input frames as well as context maps, and the warped results are fused by a synthesis network to produce the final output. This VFI approach has been fully evaluated against 12 state-of-the-art VFI methods on three commonly used test databases. The results evidently show the effectiveness of the proposed method, which offers superior interpolation performance over other state of the art algorithms, with PSNR gains up to 0.19dB.