摘 要：
随着信息技术发展，智慧校园建设成为教育信息化重要趋势。针对当前智慧校园建设中单模态感知技术导致的数据割裂、语义缺失等问题，本研究设计并优化了面向智慧校园的多模态数据融合感知网络。该网络通过分层融合架构（感知层、传输层、融合层）实现全要素感知、低时延传输和跨模态语义对齐，结合动态权重分配机制提升感知精度与鲁棒性。实验结果表明，多模态融合模型在人员定位任务中误差率较单模态降低62%，异常事件检测F1-score提升28%，同时能耗降低15%，安防响应时间缩短至30秒内。研究验证了多模态数据融合在提升校园智能化水平中的关键作用，为教育信息化转型提供了理论支撑与实践方案。
关键词：智慧校园；轻量化部署；物联网
Abstract:
With the development of information technology, the construction of smart campuses has become an important trend in educational informatization. In response to the problems such as data fragmentation and semantic loss caused by single-modal perception technology in the current construction of smart campuses, this study designs and optimizes a multi-modal data fusion perception network for smart campuses. This network achieves full-element perception, low-latency transmission and cross-modal semantic alignment through a hierarchical fusion architecture (perception layer, transmission layer, fusion layer), and improves perception accuracy and robustness by combining a dynamic weight distribution mechanism. The experimental results show that the multimodal fusion model reduces the error rate by 62% in the personnel positioning task compared with the single-modal model, increases the F1-score for abnormal event detection by 28%, reduces energy consumption by 15% at the same time, and shortens the security response time to within 30 seconds. The research verified the crucial role of multimodal data fusion in enhancing the intelligence level of campuses, providing theoretical support and practical solutions for the transformation of educational informatization.
Keywords: Smart campus; Lightweight deployment; Internet of things
--
