@article{stsimdiffbalancingspatiotemporalsimilari,
  title = {ST-SimDiff: Balancing Spatiotemporal Similarity and Difference for Efficient Video Understanding with MLLMs},
  author = {Bingjun Luo and Tony Wang and Chaoqi Chen and Xinpeng Ding},
  year = {2026},
  eprint = {2605.22158},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2605.22158},
}