Real-Time Unconscious Human Behavior Detection Incorporating Enhanced CSPDarknet53 and YOLOv9

Empirical studies focus on detecting and investigating unconscious human behavior. The study proposes to improve CSPDarknet53 with spatial splits and incorporate it with the YOLOv9 object detection model.Studies assess the usefulness of the suggested model and look at the precision and temporal complexity of recognizing unconscious human behavior. Research establishes conscious human reflection time constraints \(2500\,\text{milisecond} \approx \) and for unconscious behavior, currently depend on perception. Study have identified the importance of detecting unconscious human behavioral results of the significant impact on healthcare, secure vehicle accident prevention. The study proposed a sophisticated object detection model to investigate unconscious human activities using the UCF101 dataset with 13,320 videos spanning 101 action categories. Ten percent of the data is used for testing, while 90% of the data is used for training. YOLOv9’s improvements in accuracy led to the development of a tool for real-time behavior study. Significant result is achieved with Precision (0. 985), Recall (0. 892), F1(0.936) and mAP50 (0.980), highlighting performance in efficient and accurate monitoring of processes of the proposed model. This research explores a novel approach to leveraging YOLOv9’s advanced architecture and real-time capabilities for the specialized task of detecting unconscious human behaviors, demonstrating its practical applications and effectiveness in diverse real-world scenarios. Study associates behavioral detection with sensor capabilities and acknowledges a limited scope of comparative analysis among existing real-time object detection models.