Autoregression-free video prediction using diffusion model for mitigating error propagation
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915313491116032 |
|---|---|
| author | Ko, Woonho Park, Jin Bok Chun, Il Yong |
| author_facet | Ko, Woonho Park, Jin Bok Chun, Il Yong |
| contents | Existing long-term video prediction methods often rely on an autoregressive video prediction mechanism. However, this approach suffers from error propagation, particularly in distant future frames. To address this limitation, this paper proposes the first AutoRegression-Free (ARFree) video prediction framework using diffusion models. Different from an autoregressive video prediction mechanism, ARFree directly predicts any future frame tuples from the context frame tuple. The proposed ARFree consists of two key components: 1) a motion prediction module that predicts a future motion using motion feature extracted from the context frame tuple; 2) a training method that improves motion continuity and contextual consistency between adjacent future frame tuples. Our experiments with two benchmark datasets show that the proposed ARFree video prediction framework outperforms several state-of-the-art video prediction methods. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_22111 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Autoregression-free video prediction using diffusion model for mitigating error propagation Ko, Woonho Park, Jin Bok Chun, Il Yong Computer Vision and Pattern Recognition Existing long-term video prediction methods often rely on an autoregressive video prediction mechanism. However, this approach suffers from error propagation, particularly in distant future frames. To address this limitation, this paper proposes the first AutoRegression-Free (ARFree) video prediction framework using diffusion models. Different from an autoregressive video prediction mechanism, ARFree directly predicts any future frame tuples from the context frame tuple. The proposed ARFree consists of two key components: 1) a motion prediction module that predicts a future motion using motion feature extracted from the context frame tuple; 2) a training method that improves motion continuity and contextual consistency between adjacent future frame tuples. Our experiments with two benchmark datasets show that the proposed ARFree video prediction framework outperforms several state-of-the-art video prediction methods. |
| title | Autoregression-free video prediction using diffusion model for mitigating error propagation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.22111 |