Title: Advanced AQI interpretation using deep convolutional neural networks - a case study for Delhi

Authors: Kavita Pabreja; Gauri Banga; Gaurav Kumar Sharma; Shivansh Batra

Addresses: Lal Bahadur Shastri Institute of Management, Plot 11/07, Sector 11, Dwarka, New Delhi, 110075, India ' Accenture Solutions Pvt. Ltd, Unitech Infospace, Sector-21, Dundahera, Gurugram, Haryana, 122001, India ' Accenture Solutions Pvt. Ltd, Unitech Infospace, Sector-21, Dundahera, Gurugram, Haryana, 122001, India ' Marquee Equity, MR-1, 5th Floor, Wing-A, Statesman House, 148 Barakhamba Road, New Delhi, 110001, India

Abstract: Air pollution has a wide range of harmful impacts on public health, economy of country, and the environment. Delhi, one of the most highly polluted places globally, consistently experiences alarmingly high levels of pollution, accompanied by associated health hazards, severe economic losses due to decreased productivity, and environmental degradation. Hence, there is a need for real-time, accurate air quality index monitoring. Traditional methods, which often rely on location-based sensors, are effective but they lack broader coverage as they may not be installed at all locations or may malfunction. This research presents an advanced convolutional neural network model for AQI interpretation based on image to enable continuous and accurate air quality monitoring. An accuracy of 85.71% has been obtained for classifying AQI in six categories. To ensure practical application, a web-based platform and a chatbot has been developed, allowing users to obtain real-time AQI information through an easy-to-use interface.

Keywords: air quality index; AQI; visual geometry group; VGG16 model; deep convolutional neural networks; flask API.

DOI: 10.1504/IER.2025.147801

Interdisciplinary Environmental Review, 2025 Vol.24 No.3, pp.208 - 226

Received: 20 Sep 2024
Accepted: 03 Apr 2025
Published online: 01 Aug 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article