Human Activity Recognition and Classification Systems using YOLOv9

This research has been aimed at analyzing the efficiency of utilizing the YOLO models for determining and classifying human activities, primarily YOLOv9. The comparison of YOLO models from v1 to v9 indicates the substantial progress of used indicators for each criterion. The perfection of results is proved by the outstanding outcomes of the YOLOv9 model: $\mathbf{9 8. 2 3 \%}$ of accuracy, $\mathbf{9 8. 5 3 \%}$ of precision, $\mathbf{9 8. 6 5 \%}$ of recall, and $\mathbf{9 8. 7 5 \%}$ of F 1 rate. The measurements prove the uniqueness of the YOLOv9 model’s capacity to identify human actions properly and consistently allocate them into needed categories. The analysis demonstrates the variations of accuracy improvement in each YOLO version, which was used for the creation of YOLOv9 that underlines the model’s viability to use it for exact determination and categorization of human activities. Such advancements may play a vital role in many spheres based on the exact recognition of human behavior.