A framework of imitative behavior analysis for animal exercise courses via human pose estimation

Area of Science:

• Performance Education
• Behavioral Analysis
• Biomechanical Analysis

Background:

• Imitative behavior is crucial in performance education, particularly in Animal Exercise courses.
• Current assessment methods rely on subjective instructor judgment, leading to inconsistent evaluations.
• Objective assessment tools are needed to enhance skill development and identify deficiencies.

Purpose of the Study:

• To propose and validate a quantitative framework for evaluating imitative behavior in performance education.
• To introduce the Human Pose Estimation-Imitative Behavior Analysis (HPE-IBA) framework.
• To provide a foundation for objective, motion-based performance analysis.

Main Methods:

• Utilized a standard RGB camera to capture motion data from students and gorillas.
• Employed pose estimation with MediaPipe to extract 3D joint coordinates and calculate dynamic joint angles.
• Applied correlation analysis and two-way ANOVA to analyze imitation performance and influencing factors.

Main Results:

• The HPE-IBA framework successfully quantified imitative behavior using pose estimation.
• A statistically significant interaction between training status and gender (p<0.01) was observed in imitation performance.
• Specific joint patterns, including the right elbow and right knee, were key indicators of this interaction.

Conclusions:

• The proposed HPE-IBA framework offers an objective and standardized method for assessing imitative behavior in performance education.
• This approach enhances the application of pose estimation in acting and performance analysis.
• The framework provides a basis for broader applications in analyzing human and animal motion.