CVPR 2026 | 时空时序论文总结（天气预报，交通模拟，域自适应等）

CVPR 2026将在2026年6月5日至7日于美国科罗拉多会议中心（Colorado Convention Center）举行。

本文总结了CVPR 2026上有关时空时序的相关论文。

时空时序Topic：天气预报，交通模拟，域自适应等。

注 ：鉴于自动驾驶、空间智能及视频时空建模等领域文献多，为保持论述聚焦，本总结仅围绕天气预报 与交通模拟这两大方向展开时空数据相关工作的梳理

CVPR2026论文列表 ：++https://cvpr.thecvf.com/virtual/2026/papers.html++

1 RLFTSim: Realistic and Controllable Multi-Agent Traffic Simulation via Reinforcement Learning Fine-Tuning

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/37597++

作者：Ehsan Ahmadi, Hunter Schofield, Behzad Khamidehi, Fazel Arasteh, Jinjun Shan, Lili Mou, Dongfeng Bai, Kasra Rezaee

关键词：多智能体交通模拟，强化学习微调

2 Spatiotemporal Pyramid Flow Matching for Climate Emulation

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/37833++

arXiv ：++http://arxiv.org/abs/2512.02268++

代码 ：++https://github.com/stanfordmlgroup/spf++

作者：Jeremy Irvin, Jiaqi Han, Zikui Wang, Abdulaziz Alharbi, Yufei Zhao, Nomin-Erdene Bayarsaikhan, Daniele Visioni, Andrew Y. Ng, Duncan Watson-Parris

关键词：气候模拟，流匹配

3 ChangeBridge: Spatiotemporal Image Generation with Multimodal Controls for Remote Senisng

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/37327++

arXiv ：++https://arxiv.org/abs/2507.04678++

代码 ：++https://github.com/zhenghuizhao/ChangeBridge++

作者：Zhenghui Zhao, Chen Wu, Xiangyong Cao, Di Wang, Hongruixuan Chen, Datao Tang, Liangpei Zhang, Zhuo Zheng

关键词：时空图像生成，扩散模型，遥感

4 PFGNet: A Fully Convolutional Frequency-Guided Peripheral Gating Network for Efficient Spatiotemporal Predictive Learning

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/39501++

arXiv ：++https://arxiv.org/abs/2602.20537++

代码 ：++https://github.com/fhjdqaq/PFGNet++

作者：Xinyong Cai, Changbin Sun, Yong Wang, Hongyu Yang, Yuankai Wu

关键词：时空预测学习，全卷积框架

5 MeteorPred: A Meteorological Multimodal Large Model and Dataset for Severe Weather Event Prediction

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/36482++

arXiv ：++https://arxiv.org/abs/2508.06859++

作者：Shuo Tang, Jian Xu, Jiadong Zhang, yi chen, Qizhao Jin, Lingdong Shen, Cheng-Lin Liu, Shiming Xiang

关键词：天气事件预报，benchmark，多模态大模型

6 TrafficAlign: Aligning Large Language Models for Traffic Scenario Generation

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/37456++

代码 ：++https://github.com/TrafficComposer/TrafficAlign++

作者：Zhi Tu, Liangkun Niu, Tianyi Zhang

关键词：交通场景生成，LLM，对齐

7 STCast: Adaptive Boundary Alignment for Global and Regional Weather Forecasting

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/39880++

arXiv ：++http://arxiv.org/abs/2509.25210++

代码 ：++https://github.com/chenhao-zju/STCast++

作者：Hao Chen, Tao Han, Jie ZHANG, Song Guo, Lei Bai

关键词：天气预报

8 Towards Uncertainty-aware Unsupervised Domain Adaptation for Videos and Time-Series with Causal Optimal Transport

链接 ：++https://cvpr.thecvf.com/virtual/2026/poster/38445++

作者：Khushboo Mishra, Varun Trivedi, Tanima Dutta

关键词：无监督域自适应，最优传输，因果