Human Face Mask Recognition using ResNet 152 Model

The global health crisis, and the enforcement of mask-wearing protocols has become crucial for public safety. To automate the monitoring of these protocols, this study proposes a robust human face mask recognition system utilizing the ResNet 152 model, a deep convolutional neural network known for its accuracy in image classification tasks. The system inputs an image and processes it through successive convolutional layers- Conv 1 through Conv 4-with varying filter sizes and strides to extract detailed feature maps. These layers are followed by a max pooling operation and a fully connected layer with 1000 dimensions to synthesize the extracted features into a comprehensive representation. The output is a binary classification that determines the presence or absence of a face mask. This study implemented skip connections characteristic of ResNet architectures to enable training of deeper networks without the hindrance of vanishing gradients. Preliminary results demonstrate a high accuracy rate in face mask detection, indicating that the ResNet 152 model is a promising for enhancing public health measures through automated face mask recognition.