Human Pose Estimation And Tracking Via Keypoints Detection

Abstrak

effectiveness of the pose estimation model in accurately predicting various human poses. The consistently high mAP values across all pose classes validate the robustness and precision of the model, making it suitable for real-world applications such as motion analysis, surveillance, and sports performance tracking. Minor discrepancies in scores highlight areas for potential optimization, particularly for poses with slightly lower mAP values. overall mAP of 0.99 demonstrates that the model is highly accurate in estimating poses. Individual class scores show slight variations. For example, "L-bot-pose" has the lowest mAP (0.963).  High scores for other classes (e.g., "R-bot-pose" with 0.995) indicate that the model can detect these poses with exceptional accuracy