When Numbers Speak: Aligning Textual Numerals and Visual Instances in Text-to-Video Diffusion Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Sun, Zhengyang, Chen, Yu, Zhou, Xin, Li, Xiaofan, Chen, Xiwu, Liang, Dingkang, Bai, Xiang
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908949813395456
author Sun, Zhengyang
Chen, Yu
Zhou, Xin
Li, Xiaofan
Chen, Xiwu
Liang, Dingkang
Bai, Xiang
author_facet Sun, Zhengyang
Chen, Yu
Zhou, Xin
Li, Xiaofan
Chen, Xiwu
Liang, Dingkang
Bai, Xiang
contents Text-to-video diffusion models have enabled open-ended video synthesis, but often struggle with generating the correct number of objects specified in a prompt. We introduce NUMINA , a training-free identify-then-guide framework for improved numerical alignment. NUMINA identifies prompt-layout inconsistencies by selecting discriminative self- and cross-attention heads to derive a countable latent layout. It then refines this layout conservatively and modulates cross-attention to guide regeneration. On the introduced CountBench, NUMINA improves counting accuracy by up to 7.4% on Wan2.1-1.3B, and by 4.9% and 5.5% on 5B and 14B models, respectively. Furthermore, CLIP alignment is improved while maintaining temporal consistency. These results demonstrate that structural guidance complements seed search and prompt enhancement, offering a practical path toward count-accurate text-to-video diffusion. The code is available at https://github.com/H-EmbodVis/NUMINA.
format Preprint
id arxiv_https___arxiv_org_abs_2604_08546
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle When Numbers Speak: Aligning Textual Numerals and Visual Instances in Text-to-Video Diffusion Models
Sun, Zhengyang
Chen, Yu
Zhou, Xin
Li, Xiaofan
Chen, Xiwu
Liang, Dingkang
Bai, Xiang
Computer Vision and Pattern Recognition
Text-to-video diffusion models have enabled open-ended video synthesis, but often struggle with generating the correct number of objects specified in a prompt. We introduce NUMINA , a training-free identify-then-guide framework for improved numerical alignment. NUMINA identifies prompt-layout inconsistencies by selecting discriminative self- and cross-attention heads to derive a countable latent layout. It then refines this layout conservatively and modulates cross-attention to guide regeneration. On the introduced CountBench, NUMINA improves counting accuracy by up to 7.4% on Wan2.1-1.3B, and by 4.9% and 5.5% on 5B and 14B models, respectively. Furthermore, CLIP alignment is improved while maintaining temporal consistency. These results demonstrate that structural guidance complements seed search and prompt enhancement, offering a practical path toward count-accurate text-to-video diffusion. The code is available at https://github.com/H-EmbodVis/NUMINA.
title When Numbers Speak: Aligning Textual Numerals and Visual Instances in Text-to-Video Diffusion Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.08546