A Comparative Analysis of Deep Learning Techniques for Face Mask Detection System in Nuclear Power Plant

In high-security environments, such as the workers of a nuclear power plant, face mask detection has become extremely important in order to ensure safety and lower risks of contamination. This paper discusses the development and implementation of deep learning models with image data for detecting face masks in a nuclear power plant setting. We compare three models of great influence: CNNs, ANNs, and FNNs. All these models are trained on a large image dataset consisting of people with and without masks under various conditions of lighting and angle. This CNN-based model is known for efficiency in the field of image recognition because it has carried out several convolutional and pooling operations; it captures facial features, which results in high precision in mask detection. There are ANN and FNN models integrated in this model because its ability to optimize on the image data is less, so the comparison is done with the developed model in terms of computational tradeoff with detection performance here. Several data augmentation techniques have been made use of for noise robustness and generalization strength. Accuracy, precision, recall, and Fl score were the metrics used for the evaluation of the model performance. Conclusion it can be understood from the results obtained that with respect to accurate detection, the CNN model outperformed the ANN and FNN models. As a result, the model is quite well suited for real-time face mask detection in nuclear power plants and other similar critical infrastructures. The discovery in the safety enhancement protocols in risky environments especially shows the possibility of applying deep learning effectively for those environments.