Abnormal Human Activity Detection Using Yolov5
Keywords:
surveillance,, anomaly detection, YOLOv5,, object detection, real-time inference, model training”.Abstract
effective surveillance and anomaly detection systems are crucial in the digital age. For detecting suspicious
actions in photos and videos, this book offers a step-by-step method to implementing cutting-edge object recognition
algorithms—including YOLOv5, YOLOv6, YOLOv7, and YOLOv8 Cutting-edge its outstanding performance,
YOLOv5, renowned for its precision and quickness, is the main emphasis. Ensuring a seamless process, the guide
starts by describing the required libraries and dependencies for the project. It then addresses dataset imports, stressing
the YOLO format for annotations, and dataset preparation for efficient model training. Sample photos with bounding
boxes are visualised and class distribution examined using exploratory data analysis (EDA). The book also covers
evaluating image size and resolution compatibility with YOLOv5 criteria. The main part emphasises using the
YOLOv5 algorithm, including model loading, training, and performance evaluation measures. The trained model's
practical usesstate, such real-time or batch inference for security and surveillance objectives, are also investigated.
Although YOLOv5 is praised for its great accuracy, the book advises experimenting with later versions—including
YOLOv6, YOLOv7, and YOLOv8—for better performance
