RDTF: Resource-efficient Dual-mask Training Framework for Multi-frame Animated Sticker Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yuan, Zhiqiang, Zhang, Ting, Luo, Peixiang, Deng, Ying, Zhang, Jiapei, Jia, Zexi, Zhang, Jinchao, Zhou, Jie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918225015472128
author Yuan, Zhiqiang
Zhang, Ting
Luo, Peixiang
Deng, Ying
Zhang, Jiapei
Jia, Zexi
Zhang, Jinchao
Zhou, Jie
author_facet Yuan, Zhiqiang
Zhang, Ting
Luo, Peixiang
Deng, Ying
Zhang, Jiapei
Jia, Zexi
Zhang, Jinchao
Zhou, Jie
contents Recently, significant advancements have been achieved in video generation technology, but applying it to resource-constrained downstream tasks like multi-frame animated sticker generation (ASG) characterized by low frame rates, abstract semantics, and long tail frame length distribution-remains challenging. Parameter-efficient fine-tuning (PEFT) techniques (e.g., Adapter, LoRA) for large pre-trained models suffer from insufficient fitting ability and source-domain knowledge interference. In this paper, we propose Resource-Efficient Dual-Mask Training Framework (RDTF), a dedicated solution for multi-frame ASG task under resource constraints. We argue that training a compact model from scratch with million-level samples outperforms PEFT on large models, with RDTF realizing this via three core designs: 1) a Discrete Frame Generation Network (DFGN) optimized for low-frame-rate ASG, ensuring parameter efficiency; 2) a dual-mask based data utilization strategy to enhance the availability and diversity of limited data; 3) a difficulty-adaptive curriculum learning method that decomposes sample entropy into static and adaptive components, enabling easy-to-difficult training convergence. To provide high-quality data support for RDTFs training from scratch, we construct VSD2M-a million-level multi-modal animated sticker dataset with rich annotations (static and animated stickers, action-focused text descriptions)-filling the gap of dedicated animated data for ASG task. Experiments demonstrate that RDTF is quantitatively and qualitatively superior to state-of-the-art PEFT methods (e.g., I2V-Adapter, SimDA) on ASG tasks, verifying the feasibility of our framework under resource constraints.
format Preprint
id arxiv_https___arxiv_org_abs_2503_17735
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RDTF: Resource-efficient Dual-mask Training Framework for Multi-frame Animated Sticker Generation
Yuan, Zhiqiang
Zhang, Ting
Luo, Peixiang
Deng, Ying
Zhang, Jiapei
Jia, Zexi
Zhang, Jinchao
Zhou, Jie
Multimedia
Computer Vision and Pattern Recognition
Recently, significant advancements have been achieved in video generation technology, but applying it to resource-constrained downstream tasks like multi-frame animated sticker generation (ASG) characterized by low frame rates, abstract semantics, and long tail frame length distribution-remains challenging. Parameter-efficient fine-tuning (PEFT) techniques (e.g., Adapter, LoRA) for large pre-trained models suffer from insufficient fitting ability and source-domain knowledge interference. In this paper, we propose Resource-Efficient Dual-Mask Training Framework (RDTF), a dedicated solution for multi-frame ASG task under resource constraints. We argue that training a compact model from scratch with million-level samples outperforms PEFT on large models, with RDTF realizing this via three core designs: 1) a Discrete Frame Generation Network (DFGN) optimized for low-frame-rate ASG, ensuring parameter efficiency; 2) a dual-mask based data utilization strategy to enhance the availability and diversity of limited data; 3) a difficulty-adaptive curriculum learning method that decomposes sample entropy into static and adaptive components, enabling easy-to-difficult training convergence. To provide high-quality data support for RDTFs training from scratch, we construct VSD2M-a million-level multi-modal animated sticker dataset with rich annotations (static and animated stickers, action-focused text descriptions)-filling the gap of dedicated animated data for ASG task. Experiments demonstrate that RDTF is quantitatively and qualitatively superior to state-of-the-art PEFT methods (e.g., I2V-Adapter, SimDA) on ASG tasks, verifying the feasibility of our framework under resource constraints.
title RDTF: Resource-efficient Dual-mask Training Framework for Multi-frame Animated Sticker Generation
topic Multimedia
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.17735