The invention belongs to the technical field of
computer vision and
machine learning, and relates to a network
addiction behavior monitoring method and
system based on multi-
modal data fusion, and the method comprises the steps: firstly collecting the multi-
modal data, including physiological
signal data and behavior posture data, of a user; then, constructing a multi-element fusion judgment model based on
deep learning, inputting the physiological
signal data and the behavior posture data into the multi-element fusion model for synchronous
processing, extracting multi-
modal features and fusing the multi-modal features into a unified feature space; meanwhile, a multi-
task learning method is adopted to jointly
train the physiological
signal data and the behavior posture data; and finally, classifying and evaluating the
addiction state of the user by using a
decision tree or a
random forest algorithm. Compared with an existing network
addiction behavior monitoring method, the network addiction
behavior monitoring method has remarkable advantages in the aspects of accuracy, real-time performance,
user privacy protection, environmental adaptability,
user friendliness,
technology implementation efficiency, generalization ability, comprehensive intervention ability, end side deployment ability and the like.