Person Counting and Position Mapping in Dynamic Video Environments using YOLOv7

In order to prevent contamination in R&D labs, a person counting and bending stance detection system is proposed in this study. The model can identify and monitor up to three people while bending into positions to maintain safety and avoid contamination by examining video streams. With the Extended Efficient Layer Aggregation Network (E-ELAN) for improved computing efficiency and detection accuracy, the system is based on the YOLOv7 architecture. For model training and validation, a custom dataset annotated for person detection and bending positions was used. The model performs well across measures, including precision, recall, and mean Average Precision (mAP), according to experimental data. This means that it may be deployed with little computing expense in real-time lab situations. Future research will focus on issues like mirror reflections and further optimizing pose detection accuracy.