Building an Intelligent Home Perception System Based on Multi-Modal Information Interaction
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
In response to the problems of single interaction modality and weak perception ability in traditional smart homes, this paper proposes a multi-modal information perception Artificial Intelligence(AI) model invocation framework. It schedules visual, voice, and sensor data through natural language prompts, and combines the zero-shot visual recognition method of the cloud-based visual-language hybrid large model workflow to achieve cross-scene generalization ability without labeled training. This framework can innovatively solve the problems of heterogeneous data fusion and insufficient computing power of edge devices. Experimental results show that the multi-modal smart home perception system designed in this paper achieves an accuracy rate of over 90% in environmental perception and a precision rate as high as 92% in user intention recognition, which can provide new ideas and practical foundations for the multi-modal perception of future smart home technology.
Fields of Study & Subjects
IoT-based Smart Home Systems
Multimodal Machine Learning Applications
Context-Aware Activity Recognition Systems
Publication details
- Journal / source
- ICST Transactions on Scalable Information Systems
- Volume
- 12
- Issue
- 8
- Publication date
- 31 March 2026
- Language
- EN