Research on Laboratory Abnormal Behavior Recognition Model for Security Management Based on YOLOv5 and LSTM

Laboratory safety management is a key element in guaranteeing personnel well-being, safeguarding precious equipment, and avoiding risky accidents. Conventional safety surveillance systems in research and university laboratories mainly base their monitoring on manual monitoring or basic rule-based video monitoring, which are susceptible to human mistakes, not very scalable, and unable to effectively identify composite abnormal behaviors. Current automated systems tend to rely on static image analysis or threshold-based motion detection, leading to high rates of false alarms and limited adaptability to varied laboratory settings. To solve these problems, this research suggests a hybrid YOLOv5-LSTM-based abnormal behavior recognition model specifically for laboratory security management. YOLOv5 is utilized to perform real-time detection of staff members, laboratory equipment, and safety-critical objects, whereas LSTM networks process temporal sequences to identify normal and abnormal patterns of behavior. It facilitates the integration of both spatial and temporal comprehension of human behavior, greatly enhancing detection efficiency in intricate laboratory environments. Experimental assessment on a bespoke laboratory behavior dataset illustrates the superiority of the model over classical motion detection and standard deep learning baselines in terms of accuracy, recall, and F1-score, while achieving real-time processing rates. The findings endorse the model's viability as an effective solution for intelligent laboratory safety surveillance and pre-emptive incident prevention.