TY  - RPRT
TI  - ST-SimDiff: Balancing Spatiotemporal Similarity and Difference for Efficient Video Understanding with MLLMs
AU  - Bingjun Luo
AU  - Tony Wang
AU  - Chaoqi Chen
AU  - Xinpeng Ding
PY  - 2026
UR  - https://arxiv.org/abs/2605.22158
ID  - 2605.22158
ER  -