Change3D: Revisiting Change Detection and Captioning from A Video Modeling Perspective (CVPR 2025 Highlight 🔥🔥🔥)
**_A simple and efficient framework for change detection and captioning tasks._**
[Duowang Zhu](https://scholar.google.com/citations?user=9qk9xhoAAAAJ&hl=en)
1, [Xiaohu Huang](https://scholar.google.com/citations?user=sBjFwuQAAAAJ&hl=en)
2, [Haiyan Huang](https://www.researchgate.net/profile/Haiyan-Huang-11)
1, [Hao Zhou](https://scholar.google.com/citations?user=xZ-0R3cAAAAJ&hl=zh-CN)
3, and [Zhenfeng Shao](http://www.lmars.whu.edu.cn/prof_web/shaozhenfeng/index.html)
1*
1 Wuhan University
2 The University of Hong Kong
3 Bytedance
[

](https://openaccess.thecvf.com/content/CVPR2025/papers/Zhu_Change3D_Revisiting_Change_Detection_and_Captioning_from_A_Video_Modeling_CVPR_2025_paper.pdf)
[

](https://arxiv.org/pdf/2503.18803)
[

](https://zhuduowang.github.io/Change3D)