Air pollution is a concern to the health of all living beings. It is essential to check on the quality of air in the surroundings. This article presents an IoT-based real-time air quality index (AQI) estimation technique using images and weather sensors on Indian rods. A mixture of image features, i.e., traffic density, visibility, and sensor features, i.e., temperature and humidity, were used to predict the AQI. Object detection and localization-based Deep Learning (DL) method along with image processing techniques were used to extract image features while an Machine Learning (ML) model was trained on those features to estimate the AQI. In order to conduct this experiment, a dataset containing 5048 images along with co-located AQI values across different seasons was collected by driving on the roads of Hyderabad city in India. The experimental results report an overall accuracy of 82% for AQI prediction.
- An IoT based novel methodology is proposed to estimate the real-time AQI into five levels using traffic images and weather parameters.
- To the best of authors’ knowledge, this work is the first of its kind to achieve this on Indian roads.
- An entirely new traffic dataset is collected on Indian roads containing 5048 images and related weather data with co-located ground truth PM values. The dataset contains samples across the Indian city of Hyderabad in different seasons.
- The proposed method achieved overall 82% accuracy considering PM variation due to season.
To clone this repository,
git clone omi-kron/IoT-based-AQI-Estimation-using-Image-Processing-and-Learning-MethodsAfter that run if you are using pip,
pip install -r requirements.txtand in case of conda,
conda env create -f environment.yml
After creating the environment, you can activate it with:
conda activate my_project_envThe proposed methodology for the IoT-based AQI Estimation using Image Processing and Learning Methods was executed through the following steps:
Before commencing data collection, a robust hardware setup was assembled. This setup consisted of A Raspberry Pi 3B+ (Rpi 3B+) microcontroller unit (MCU) and a PiCamera are connected to it to capture and process the vehicle images. The other sensors that were interfaced with the MCU include BME280 for temperature and humidity. A SDS011 Nova PM sensor was used for measuring the PM2.5 and PM10 concentrations.
With the help of above hardware setup, the data collection device is prepared. The data collection device is mounted on the top of the vehicle and roamed around the city of hyderabad.
The above device is made of Raspberry Pi 3B+ and extensively runs on the Python3. One needs to run following command on the Rpi to start data collection.
python3 data_collection_rpi.pyTo run this file on boot one needs to create a service file. To create a service file, automation.md can be followed.
automation.md
The GPS coordinates are collected using the android phone with the application named GPS Logger installed in it: GPS Logger.
The sensor data which is collected using the data collection device is need to be calibrated and pre-processed. The pre-processing involves estimating the AQI from the PM2.5 and PM10 values which are collected using the sensor. The preprocessing is done in the ipython notebook file called,
sensor_data_preprocessing.ipynb
To detect the vehicles from a given image, You-Only-Look-Once version 5 (YOLOv5) was trained on Indian Driving Dataset (IDD).
See the YOLOv5 Docs for full documentation on training, testing and deployment. See below for quickstart examples.
Install Clone repo and install requirements.txt in a Python>=3.8.0 environment, including PyTorch>=1.8.
git clone https://github.com/ultralytics/yolov5 # clone
cd yolov5
pip install -r requirements.txt # install
The final dataset was prepared using concatenation of sensor features and image features. This was acheived using python and dataset is also cleaned as there are few outliers in it, that might affect the accuracy of the overall accuracy of the classification model.
For a comprehensive understanding of the conducted experiments and the outcomes achieved, I encourage you to refer to the accompanying paper. In particular, you can gain detailed insights into all the experiments and their corresponding results by reviewing the content presented in the IPython notebook titled "main_script.ipynb". This notebook serves as a comprehensive resource, providing thorough explanations and analyses of each experiment, making it an invaluable reference to delve into the research's intricacies and findings.
main_script.ipynb
@INPROCEEDINGS{10152272,
author={Nitin Nilesh, Ishan Patwardhan, Jayati Narang and Sachin Chaudhari},
booktitle={2022 IEEE 8th World Forum on Internet of Things (WF-IoT)},
title={IoT-based AQI Estimation using Image Processing and Learning Methods},
year={2022},
volume={},
number={},
pages={1-5},
doi={10.1109/WF-IoT54382.2022.10152272}}
The code, platform, and dataset are made available for academic research purpose only.
This research was supported partly by National Geospatial Programme (NGP), India, under grant no. 2073 (2020), PRIF Social Incubator Program (2019) and the Ministry of Electronics and Information Technology (MEITY), Govt. of India under grant no. 3070665 (2020), with no conflict of interests.
Email - Nitin Nilesh, Om Kathalkar