MAS-LLaVA: Motion-Aware Adaptive Sampling for Training-Free Video Large Language Models
Presented at IEEE International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA), 2026
MAS-LLaVA: Motion-Aware Adaptive Sampling for Training-Free Video Large Language Models
Published at IEEE International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA), 2026.
MAS-LLaVA introduces motion-aware token and frame sampling for efficient training-free video large language model inference.
Recommended citation: Tang, Jialin; Bai, Yu. (2026). “MAS-LLaVA: Motion-Aware Adaptive Sampling for Training-Free Video Large Language Models.” IEEE International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA).
Recommended citation: Tang, Jialin; Bai, Yu. (2026). "MAS-LLaVA: Motion-Aware Adaptive Sampling for Training-Free Video Large Language Models." IEEE International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA).
Download Paper