Verified65% confidenceFactExact time
FVD was proposed by Unterthiner et al. in 2019; it uses an I3D network pretrained on the Kinetics dataset to extract spatiotemporal video features and computes the Fréchet distance between the distributions of real and generated videos.
3
Sources
65%
Confidence
Long-term
Relevance
7/13/2026
First Seen
Sources
欧拉运动引导：解决图像动画长序列漂移的新方案
redditr/learnmachinelearning7/12/2026
Related Claims
UnverifiedFVD是FID在视频领域的扩展版本，在计算分布差异时考虑了时间维度的特征72% similarUnverifiedCLIP score and FVD (Fréchet Video Distance) are quantitative metrics that can be used to evaluate overall image quality.70% similarUnverifiedCurrent mainstream video processing uses a static approach in which the model ingests video content frame by frame at a fixed frame rate (defaulting to 1 FPS, adjustable via API).69% similarUnverified学界正在探索基于NeRF（神经辐射场）和3DGS（三维高斯溅射）的视频生成方案，有望解决跨帧空间一致性问题67% similarUnverifiedIn recent years, deep learning-based matting solutions such as MODNet and RVM (Robust Video Matting) have achieved pixel-level precise separation67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/501375API
curl https://kongchang.com/api/v1/knowledge/claims/501375MCP
get_claim(id=501375)