Photo-Based Monitoring of Particulate Matter in the Campus: A New Strategy for Student Health

Abstract

Abstract It is imperative for the students’ future health to ensure the students in good physical levels. Recent years have witnessed the increasingly serious harm to student health caused by the continually growing concentration of Particulate Matters (PMs). Consequently, the task of preventing and controlling PM concentrations in the campus is eagerly required. A well-designed model for the monitoring of PM (as the basis for PM prevention and control) has posed a big challenge. Prior works have revealed that photo-based methods are available for the monitoring of PM. Towards validating the effectiveness of existing methods for PM monitoring in the campus, we construct a novel dataset that involves 1,500 photos collected in the Beijing University of Technology. Results confirm that stated-of-the-art methods are far from ideal for the monitoring of PM in the campus. To solve the aforesaid issue, this paper further proposes a novel photo-based PM monitoring model by using the weighted average method solved by LASSO regression to fuse the above methods’ outputs tested to infer the PM values. Results demonstrate the superiority of our proposed model as compared to state-of-the-art methods on the large-scale AQPDBJUT dataset.