Search

Coronavirus disease 2019 has seriously affected the world. One major protective measure for individuals is to wear masks in public areas. Several regions applied a compulsory mask-wearing rule in public areas to prevent transmission of the virus. Few research studies have examined automatic face mask detection based on image analysis. In this paper, we propose a deep learning based single-shot light-weight face mask detector to meet the low computational requirements for embedded systems, as well as achieve high performance. To cope with the low feature extraction capability caused by the light-weight model, we propose two novel methods to enhance the model’s feature extraction process. First, to extract rich context information and focus on crucial face mask related regions, we propose a novel residual context attention module. Second, to learn more discriminating features for faces with and without masks, we introduce a novel auxiliary task using synthesized Gaussian heat map regression. Ablation studies show that these methods can considerably boost the feature extraction ability and thus increase the final detection performance. Comparison with other models shows that the proposed model achieves state-of-the-art results on two public datasets, the AIZOO and Moxa3K face mask datasets. In particular, compared with another light-weight you only look once version 3 tiny model, the mean average precision of our model is 1.7% higher on the AIZOO dataset, and 10.47% higher on the Moxa3K dataset. Therefore, the proposed model has a high potential to contribute to public health care and fight against the coronavirus disease 2019 pandemic.

Facial expression recognition has been applied widely in human-machine interactions, security and business applications. The aim of facial expression recognition is to classify human expressions from their face images. In this work, we propose a novel neural network-based pipeline for facial expression recognition, Hybrid Separable Convolutional Inception Residual Network, using transfer learning with Inception residual network and depth-wise separable convolution. Specifically, our method uses multi-task convolutional neural network for face detection, then modifies the last two blocks of the original Inception residual network using depthwise separable convolution to reduce the computation cost, and finally utilizes transfer learning to take advantages of the transferable weights from a large face recognition dataset. Experimental result on three different databases - the Radboud Faces Database, Compounded Facial Expression of Emotions Database, and Real-word Affective Face Database, shows superior performance compared with the existing studies. Moreover, the proposed method is computationally efficient and reduces the trainable parameters by approximately 25% than the original Inception residual network.

Travel time estimation plays a core role in intelligent traffic systems, as people or smart vehicles can choose a better travel route in advance, while local authorities can make urban traffic planning based on it. The paper proposes a dual traffic network topology coupled with improved delay models to estimate travel time in urban traffic system. The structure of the system is demonstrated by the dual traffic network (DTN) topology. It hires nodes to representjoint points of links and intersections; it also has two kinds of edges-links and turning cases of intersections. The BPR model is improved by concerning the influence of pedestrians and the lane width, and the HCM model is improved by adding travel time at intersections and considering three different turning cases to match with the DTN topology. The proposed methodology is evaluated by a comparison and an experiment in Changzhou, China. The error is greatly reduced, even though the degree of saturation exceeds 0.72. This proposed estimation methodology can improve the estimation accuracy significantly, especially in high degree of saturation cases.