Human phone usage recognition based on OpenPose
Yuyan Li, Jiawen Luo, Jinghao Wen, Yangjia Zhang · 2022 3rd International Conference on Computer Vision, Image and Deep Learning & International Conference on Computer Engineering and Applications (CVIDL & ICCEA) · 2022
With the development of science and technology, more and more people use mobile phones outdoors, which will cause danger to pedestrians using mobile phones and drivers on the road. This paper uses OpenPose method to detect whether pedestrians on the road are using mobile phones. We describe an approach to recognize and classify pedestrian posture in an individual context, more precisely in open door environment. The posture belongs into two main groups: phone usage and not. This paper uses 15 points OpenPose model to estimate human key points. Then select and calculate features using the estimated coordinates. Four neural network models are trained and tested using different feature combinations to classify the posture. The best model achieves 89.66% accuracy in classification.