Jiaqi Li

2026-09-01 — 2029-07-01
👤 Personal Information
- Class of 2026, Master candidate in Electronic Information
- Document Analysis and Pattern Recognition Laboratory, School of Information Science and Technology, Shihezi University
🔬 Research Interests
- Remote‑Sensing Change Detection Based on Large‑Model and World‑Model
In the training of remote‑sensing change‑detection models, high‑quality real‑world change samples are difficult to obtain and expensive to annotate. To address this issue, the diffusion model is introduced as a sample generation module. Based on existing real remote‑sensing images, a large number of synthetic change samples are generated to expand the training dataset. Combined with the analytical capabilities of large‑models and world‑models, an appropriate change‑detection discrimination paradigm is adopted to mine inherent feature differences between real ground‑object changes and synthetic generated changes. A robust remote‑sensing change‑detection system with strong anti‑interference ability and excellent generalization performance is finally constructed.
Keywords: remote‑sensing change detection, diffusion model, sample generation, large model, world model, dataset augmentation
📄 Academic Outputs
None
🏆 Honors & Awards
None
