Air Pollution Detection using Resnet-50

Abstract

Pollution has become the major concerning problem these days. Many cities are being affected by the pollution. Many people are suffering from various diseases because of the pollution around them. There is an existing system to monitor the air quality, but it consists of monitoring stations where hardware components like sensors are being used, which is more complex and expensive. Through this proposed model people can monitor the Particulate Matter 10 in their location by capturing the images of the surrounding area. In this approach particulate matter 10 is predicted by taking image as input. The images are labeled with the correct Air Quality Index (AQI) value and are extracted using regular expression. For estimating the Air Quality Index value, the convolutional neural network models ResNet-50 and mobileNet are trained. The accuracy rate of the models is determined at the end. ResNet-50 yields high accuracy rate compared to mobileNet.