MEMOS : Multimodal Educational Mentor and Optimisation System Based on Multi-Agent
Yaodong Hu, Zhen Chen, Yuxiang Lin, Junyu Wang, Yishan Liu, Weiran Lin, Lie Zhang, Min Guo · 2024
In the rapidly evolving field of AI, education stands as a significant application domain. This paper presents an intelligent system designed to assist both teaching and learning by integrating multimodal large models and a multi-agent framework. Our system leverages intelligent hardware to record real classroom videos, which are then processed by multimodal models for recognition, comprehension, and description. These courses’ data are stored in a vector database and interacted with large models to automatically generate teaching and learning reports. The system's core functions include automated lecture recording, note generation, and interactive feedback mechanisms, fostering an adaptive and efficient educational environment. This process aims to enhance teaching content and planning for educators, while also improving learning efficiency for students.