Research and development of intelligent camera-based safety monitoring and alert software for students in laboratories at Vietnam Maritime University

The study focuses on developing safety monitoring and alert software for students in practice laboratories at Vietnam Maritime University, utilizing security cameras integrated with artificial intelligence (AI). The YOLOv8n deep learning model was chosen to detect violations such as not wearing proper uniforms, unsafe behaviors, and fire hazards. The research team collected and labeled 3,700 images and trained the model on the Google Colab platform over 200 epochs, achieving an average precision (mAP) of nearly 70%. The software displays real-time data from the camera and integrates a Flask API to send alerts and images to a management web interface. Experimental results demonstrate that the system is capable of accurate recognition, rapid response, and stable operation, effectively supporting safety monitoring and assessing students’ compliance with safety regulations in practice.