Public Policy and Big Data Algorithm Strategies for Enhancing Tobacco Control Compliance in Public Places: Practice and Effectiveness Evaluation

In order to assist the efficiency of smoking detection for tobacco control in public places in a high-speed and accurate manner, an improved MTCNN-based face detection algorithm is used to reduce the interference of non-face regions by locating the face region in the video frame image. Smoking behavior features in the mouth region are also extracted to improve the accuracy of smoking behavior detection. Background noise of candidate regions is filtered using EfficientDet network to classify foreground objects and candidate regions with stronger features to accurately recognize smoking behavior. Targeted legislative strategy measures are proposed to strengthen tobacco control legislation in key areas, categorize and manage different groups of smokers, improve tobacco control facilities in public areas, and increase law enforcement and public participation. Validation found that the method in this paper can effectively identify smoking behavior, and the accuracy range of multi-person smoking action identification is up to 98.33%. The distance trajectory of smokers fluctuates in the range of 19.5-116.3. The number of smokers in the area before and after the detection of this paper's strategy is reduced by 821, and the public attitude support rate of tobacco control compliance is as high as 90.36%, and the legislative strategy based on big data algorithm can be popularized and applied in the field of tobacco control.