@misc{indiciae86d681367969,
  title = {Semantic-Geometric Dual Compression: Training-Free Visual Token Reduction for Ultra-High-Resolution Remote Sensing Understanding},
  author = {Yueying Li and Fengxiang Wang and Yan Li and Mingshuo Chen and Mengying Zhao and Long Lan},
  year = {2026},
  url = {https://arxiv.org/abs/2604.11122},
  note = {Source identifier: 2604.11122}
}