DGFO: Dynamic Guidance Fit Orthodontics Video Generation

Read full text Summarise with VIRAT
Authors: Duoyang Liao, Xuran He, Junfeng Xiong, Tao He, Ruonan Sun, Qixiong Yao
Cited by: 0
References: 29

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

In recent years, the rise of Text-to-Video (T2V) large model has facilitated people to generate expected videos according to their own ideas. With deeper research and training, these large models have considerable application prospect in many aspects. In the medical image and video generation, T2V models can generate more vivid videos, providing more intuitive medical effects for the therapists and patients. But the fact is that these methods now don’t perform well in the amplitude of the teeth movement in the generated video for orthodontics. In this paper, to bridge the gap between AI image generation and orthodontic treatment, we introduce our DGFO (dynamic guidance fit orthodontic video generation) modules, the dynamic guidance adapter that fits orthodontic video generation, which use mixed functions to control the classifier guidance change by the timestep of denoising loop. Our approach allows orthodontic features to be characterised in video generation by using phased thoughts. We trained base large T2V models with oral images, and the experiments shows that our method could have better performance on indicators like temporal consistency, and so on.

Fields of Study & Subjects

Generative Adversarial Networks and Image Synthesis Orthodontics and Dentofacial Orthopedics Multimodal Machine Learning Applications

Publication details

Journal / source
IEEE Access
Volume
14
Pages
54625–54637
Publication date
1 January 2026
Language
EN