A Survey of Multimodal Large Language Model from A Data-centric
Perspective
A Survey of Multimodal Large Language Model from A Data-centric
Perspective
Human beings perceive the world through diverse senses such as sight, smell, hearing, and touch. Similarly, multimodal large language models (MLLMs) enhance the capabilities of traditional large language models by integrating and processing data from multiple modalities including text, vision, audio, video, and 3D environments. Data plays a pivotal role …