PM2:5 monitoring using images from smartphones in participatory sensing

The Sensable City: A Survey on the Deployment and Management for Smart City Monitoring

“…Many applications have been built or designed based on crowd sensing. For example, to estimate the air quality and PM2.5 in cities, the study in [47] has proposed to use the photo taken by smart phones and tagged with GPS data. The MIT VTrack [48] and Mobile Millennium project [49] proposed to use smart phones to provide better traffic information, such as finer-grained traffic status and accurate travel time estimations.…”

“…), and geography data. Besides, a learning-based method has been proposed in [47] to extract air quality data from images taken by smart phones. Inspired by this, the camera readings taken by vehicles could greatly help in air quality monitoring in the near future.…”

ResNet-LSTM for Real-Time PM2.5 and PM₁₀ Estimation Using Sequential Smartphone Images

“…Zhang et al [8] made good use of multikernel learning to estimate air quality. Liu et al [9] adopted similar features and used a linear least square regression to estimate PM 2.5 values via smartphone-taken images. Instead of using basic image features, Gu et al [10] constructed a picture-based predictor.…”

“…To overcome such challenge, previous estimation studies designed and used feature-based machine learning models [7], [8], [9], [10], [11], such as Support Vector Regression (SVR), to estimate PM 2.5 values, capitalizing on features manually selected from images (such as contrast and saturation). However, as these feature-based models are highly dependent on how features are constructed, their performance can be easily distorted by any change in environmental conditions.…”

MCS-RF: mobile crowdsensing–based air quality estimation with random forest

“…We use the radiometric calibration method to recover the image. 29 Mobile phone photos usually contain two parts: the sky and the scene. These two parts are quite different in image features.…”