지민몬
[CVPR 2024] MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding