songweige/content-debiased-fvd
[CVPR 2024] On the Content Bias in Fréchet Video Distance
This helps researchers and practitioners evaluate the realism and quality of videos generated by AI models. It takes a collection of generated videos and compares them to a set of real videos, providing a Fréchet Video Distance (FVD) score that more accurately reflects motion quality. This tool is for anyone developing or assessing video generation AI models, such as those in computer vision research or content creation.
144 stars. No commits in the last 6 months.
Use this if you need a more reliable and less 'content-biased' method to quantify how realistic your AI-generated videos are, particularly regarding their motion.
Not ideal if you are looking for a qualitative assessment or subjective user feedback on video quality, as this provides a quantitative metric.
Stars
144
Forks
9
Language
Python
License
MIT
Category
Last pushed
Sep 28, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/songweige/content-debiased-fvd"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
hao-ai-lab/FastVideo
A unified inference and post-training framework for accelerated video generation.
ModelTC/LightX2V
Light Image Video Generation Inference Framework
thu-ml/TurboDiffusion
TurboDiffusion: 100–200× Acceleration for Video Diffusion Models
PKU-YuanGroup/Helios
Helios: Real Real-Time Long Video Generation Model
PKU-YuanGroup/MagicTime
[TPAMI 2025🔥] MagicTime: Time-lapse Video Generation Models as Metamorphic Simulators