Multi-feature fusion model for interactive behaviour recognition in university classrooms using convolutional neural networks and temporal attention mechanisms

Key Points

• To develop a multi-feature fusion model for recognizing interactive behaviors in university classrooms using neural networks.
• Implemented a convolutional neural network to analyze video data
• Employed temporal attention mechanisms to enhance feature recognition
• Evaluated the model's performance in real classroom settings
• Achieved a significant improvement in recognition accuracy over traditional methods
• Demonstrated enhanced performance metrics with p-value < 0.05
• Provided insights into the effectiveness of temporal attention in behavior recognition

Abstract