The recognition of human activity is crucial as the Internet of Things (IoT) progresses toward future smart homes. Wi-Fi-based motion-recognition stands out due to its non-contact nature and widespread applicability. However, the channel state information (CSI) related to human movement in indoor environments changes with the direction of movement, which poses challenges for existing Wi-Fi movement-recognition methods. These challenges include limited directions of movement that can be detected, short detection distances, and inaccurate feature extraction, all of which significantly constrain the wide-scale application of Wi-Fi action-recognition. To address this issue, we propose a direction-independent CSI fusion and sharing model named CSI-F, one which combines Convolutional Neural Networks (CNN) and Gated Recurrent Units (GRU). Specifically, we have introduced a series of signal-processing techniques that utilize antenna diversity to eliminate random phase shifts, thereby removing noise influences unrelated to motion information. Later, by amplifying the Doppler frequency shift effect through cyclic actions and generating a spectrogram, we further enhance the impact of actions on CSI. To demonstrate the effectiveness of this method, we conducted experiments on datasets collected in natural environments. We confirmed that the superposition of periodic actions on CSI can improve the accuracy of the process. CSI-F can achieve higher recognition accuracy compared with other methods and a monitoring coverage of up to 6 m.
Teacher–student interaction is central to classroom learning, yet traditional observation and machine-learning approaches often remain inefficient and subjective. This study explores the use of multimodal large language models (MLLMs) for systematic analysis of classroom dynamics. We fine-tuned VisualGLM-6B on 2,380 annotated images from 30 classroom videos, covering five interaction types: guided, collaborative, questioning, independent, and exhibitive. LoRA-based fine-tuning combined with prompt engineering was employed to enhance interpretability and domain-specific accuracy. Model performance was assessed through confusion matrices, BERTScore, and expert comparisons. The fine-tuned model achieved 82% overall accuracy, performing best on guided, independent, and exhibitive interactions, while collaborative and questioning types remained challenging. Compared with expert annotation, the model provided more structured and interpretable outputs, though occasional misclassifications and hallucinations persisted. These findings demonstrate the feasibility of applying MLLMs for efficient, objective analysis of teacher–student interactions and highlight future directions such as incorporating audio inputs and larger datasets to further advance educational research methodologies.