A teacher-student interaction intention recognition and response system based on end-side semantic understanding

Abstract

The application provides a teacher-student interaction intention recognition and response system based on end-side semantic understanding, and relates to the field of semantic understanding systems. The teacher-student interaction intention recognition and response system based on end-side semantic understanding comprises the following steps: S1, pre-importing student identity information and teacher voiceprint model. The teacher-student interaction intention recognition and response system based on end-side semantic understanding deploys a lightweight semantic understanding model on an end-side device, combines a cache and a dynamic confidence threshold, realizes high-robustness teacher-student interaction intention recognition, realizes low-delay (≤500 ms) student close-up and recording that meet the real-time interaction requirements of a classroom through digital cropping technology, fuses visual and audio multi-modal information, introduces time and intensity constraints, and improves the accuracy of student response behavior judgment, the whole system independently runs on the end side, all data are locally encrypted and stored, the privacy and safety of educational data are ensured, and the system has abnormal degradation capability.

Images