AI Paper Atlas

Research map / Video / Video generation

FETV: A Benchmark for Fine-Grained Evaluation of Open-Domain Text-to-Video Generation

Yuanxin Liu, Lei Li, Shuhuai Ren, Rundong Gao, Shicheng Li, Sishuo Chen, Xu Sun, Lu Hou

NeurIPS 2023 · Video › Video generation · approach: Compositional text-to-video diffusion

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fine-tune CLIP on MSR-VTT video-text retrieval (CLIPScore-ft)CLIPScore · FETV video-text alignment, all categoriesKendall tau_c: 0.177 -> 0.224Improved
replace CLIP with BLIP (BLIPScore)CLIPScore · FETV video-text alignment, all categoriesKendall tau_c: 0.177 -> 0.235Improved
replace CLIP with video-pretrained UMT (UMTScore)CLIPScore · FETV video-text alignment, all categoriesKendall tau_c: 0.177 -> 0.309Improved
replace CLIP with UMT (UMTScore)BLIPScore · FETV video-text alignment, all categoriesKendall tau_c: 0.235 -> 0.309Improved
use Video LLM Otter with yes/no VQA instead of CLIP similarityCLIPScore · FETV video-text alignment, all categoriesKendall tau_c: 0.177 -> 0.081Hurt
UMTScore Quantity vs Color categoriesUMTScore · FETV video-text alignment, all categoriesKendall tau_c: 0.394 vs 0.304Mixed
replace I3D encoder in FVD with UMT vision encoder (FVD-UMT)FVD · FETV video-text alignment, all categoriesranking consistency with human qualityImproved
increase number of generated videos for FID/FVD (ablation)FID/FVD · FETV video-text alignment, all categoriesFID/FVDMixed
zero-shot ZeroScope trained from ModelScopeT2V vs original ModelScopeT2VZeroScope · FETV video-text alignment, all categorieshuman alignmentHurt
prompt complexity simple to complex (ablation)T2V models · FETV video-text alignment, all categorieshuman alignmentNo clear effect
use fine-grained vs overall alignment ratingmanual evaluation · FETV video-text alignment, all categorieshuman alignmentNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy