Urban air pollution is a significant environmental concern that affects public health and quality of life. Monitoring and classification of air pollution levels are essential for implementing effective mitigation strategies and safeguarding human health. In this study, we present a novel approach to classify urban air pollution into six distinct categories using convolutional neural networks (CNNs).
In the study, a data set consisting of 600 images in total was used, by randomly taking 100 images for each category from 12000 image data consisting of air quality images of two different cities. There are a total of six air pollution classes in the data set: "good", "moderate", "unhealthy for sensitive groups", "unhealthy", "very unhealthy" and "hazardous". These classes are categorized according to air quality index (AQI).
We design and train a CNN architecture tailored for multi-class classification tasks, leveraging transfer learning and data augmentation techniques to enhance model performance. The CNN is trained on a large dataset of labeled air quality samples, enabling it to learn intricate patterns and correlations between different pollutant levels.
Experimental results demonstrate the efficacy of the proposed CNN-based approach in accurately categorizing urban air pollution into six predefined classes. The trained model achieves an impressive accuracy rate of 97% on the testing dataset, indicating its robustness and reliability in detecting polluted air.
Our study contributes to the advancement of air quality monitoring and management systems, providing a valuable tool for local authorities and policymakers to assess pollution levels, identify hotspots, and implement targeted interventions to improve urban air quality and protect public health.
ORCID NO: 0000-0001-9105-508X
Anahtar Kelimeler: CNN, Transfer Learning, Deep Learning, Air Pollution
|
