3 results about "Behavioural state" patented technology

This application provides a game behavior planning method, apparatus, computer device, and medium. The method includes: acquiring the current state information and target task planning of a non-player character in the game, wherein the current state information includes the current behavior and the current behavior state, and the target task planning includes the target behavior state; quantifying the current behavior state and the target behavior state, and calculating the distance between the quantified current behavior state and the target behavior state; acquiring behavior route information from an action library based on the distance, using the current behavior as a reference; and performing behavior planning on the non-player character based on the behavior route information, so that the non-player character completes the target task planning. The method proposed in this application can acquire the behavior route of the non-player character according to a given target task planning, and then control it to complete the target planning. This method improves the development speed of the non-player character, and the behavior range of the non-player character is wide, which can improve the player's user experience.

The invention discloses a student management system and method based on classroom behavior recognition, and belongs to the field of education and teaching, and the system obtains classroom video data based on the multi-angle visual collection and microphone array signal collection technology, obtains classroom audio signal data at the same time, and achieves the student management based on target detection, key point posture recognition and time sequence behavior recognition. The method comprises the steps of extracting spatial trajectory and behavior state information of students in a classroom, obtaining audio feature parameters of student emotion activeness and classroom participation degree based on voice signal framing analysis, voice signal feature extraction and a deep emotion recognition network, and obtaining association between the students and teaching based on audio and video feature fusion and a student behavior right reconstruction model. Student groups are subjected to grouping analysis through a clustering algorithm, a teaching mode is recommended in a personalized manner based on attraction characteristics, and real-time evaluation and intelligent improvement of classroom teaching quality are realized.