Hazard identification of college chemical laboratory based on image recognition

In recent years, computer vision technology has advanced rapidly, particularly in the area of intelligent analysis using image recognition. This paper proposes an image recognition algorithm based on the YOLOv5 model, which aims to improve laboratory safety hazard detection. The system operates by pre-labeling images of laboratory equipment and conducting targeted training, which is then combined with a Python-based judgment code to identify various items and facilities within hazardous chemical laboratories, such as reagent cabinets, gas cylinders, and electric heating plates. The algorithm compares the real-time recognition results with a predefined "normal" state to detect anomalies, including malfunctioning devices or missing components. When discrepancies are identified, the system automatically generates a hazard detection report. This approach significantly reduces the reliance on the subjective experience of laboratory managers, offering a more efficient, objective, and automated solution for hazard detection. As a result, it has substantially improved the overall efficiency of laboratory hazard investigation.