Air pollution in metropolitan cities such as in India has significantly increased due to fast urbanization and

consistently rising number of vehicles along with its motor activity. Pollutants such as Nitrogen Oxides (NOx),

Sulphur Dioxide (SO₂), Particulate Matter (PM₂.₅, PM₁₀), Hydrocarbons (HC), Carbon Monoxide (CO) and

others are important components emitted by the vehicle which have negative impact on the environment and

human health. Precise emission inventory estimation is crucial for developing management plans and policies

information are extracted from traffic images and video using deep learning models such as Convolutional

Studying this is done using a spatial grid-based approach to analysing the dispersion of emissions within urban

areas and temporal variations over hourly, daily and seasonal scales. Their comparative study with the emissions

produced at Mumbai shows that there are significant variations in emissions because of the population density,

the traffic composition and the urban infrastructure. Incorporated with AI, this emission inventory model with

enhanced accuracy and efficiency offers significant improvement from the typical approach to valuable insights

Emission inventories are a fundamental tool to provide information about the amount of pollutant emitted by

various sources and their distribution in space and time. The basic statistical models and traffic activity data are

the main input into traditional emission inventory models. However, these approaches are rather limited in their

temporal resolution, or are not as confident, notwithstanding suboptimal traffic data quality, variability of vehicle

New opportunities for improving emission inventory development include recent developments in Artificial

Intelligence (AI), machine learning (ML), computer vision, and geographic information systems (GIS). AI-

driven methods are capable of automatically extracting traffic data from photos and videos, enhances the

quality problem of Delhi. They studied the pollution levels of air before and after CNG & observed significant

reduction in pollution levels like particle matter and taking down other negative emissions as a result of using

CNG. The study emphasized the importance of using cleaner fuel technologies in order to reduce the levels of

Gurjar et al. (2010) presented research on air pollution in major cities throughout the world and how this will be

affecting human health. The research exposed the importance of the contribution of the transportation sources to

air pollution in the urban context and the impacts of this in terms of health issues. The authors suggest that

developing an emission inventory is an important step in developing pollution abatement programs and that the

Nagpure et al (2011) studied the correlation between the air quality deterioration, rapid urbanisation & growth

humidity in relation to emission levels and pollutant dispersion. The authors emphasized that the meteorological

conditions need to be taken into consideration when preparing emission inventory models, to consider the

Singh and Pandey (2013) have analysed the air pollution in the cities as a spatial analysis using GIS. They have

already demonstrated effectiveness of GIS solutions in the identification of pollution hot spots and knowledge

of the spatial distribution of emissions. The study offered important insights into the use of GIS for integrating

The On-road vehicular emission inventory was developed for Lucknow city, India by Shukla & Alam (2014),

using GIS. By using traffic activity data, the vehicle categories and the emission factors, the pollutant emission

at certain locations was estimated. The results demonstrated the usefulness of the GIS tools to generate detailed

LeCun, Bengio and Hinton (2015) gave an advance survey of deep learning and its application to pattern

the model attained a very high level of accuracy of detection and reacts very quickly, making it very compatible

with the monitoring system developed for the traffic. Ever since that time, YOLO has become one of the best

For time-series prediction and sequential data analysis, Hochreiter and Schmidhuber (2017) talked about a

special type of Recurrent Neural Networks (RNN), the Long short term memory networks (LSTM). They also

played an important role in developing a scientific basis for forecasting traffic flow and future emissions from

Kumar et al. (2018) has given an extensive overview of the growing applications of cheap, sensor technologies

where air pollution measurements were being made in the urban setting. The study highlighted the value of

continuous measurement of the environment and the value of real-time data collection. The authors pointed out

Beig et al. (2020) created a detailed emission inventory of anthropogenic PM₂.₅ and PM₁₀ emissions for India.

Sahu et al. (2020) have developed a comprehensive emission inventory anthropogenic PM₂.₅ and PM₁₀ emissions

for India. They found major emission sources to be transportation, industrial activities and energy generation.

The study provided an excellent inventory database which could be used for air quality modelling development

Guttikunda, Nishadh and Jawahar (2021) assessed Air Pollution Knowledge and characteristics of emission in

Mangaraj et al. (2022) developed and analysed a comprehensive EI for the city of Bengaluru. The research

utilized the activity data, emission factors and spatial analysis methods for the quantification of urban air

pollutant emissions. The city-carbon accounting studies emphasized the need for city level emission inventories

Sahu et al. (2023) have studied an increase in air pollutant emission in Delhi over the decade period. With the

fast urbanization, population and transport growth, they discovered that the emission load has a good magnitude.

The research highlighted the need for regular monitoring, and upkeep of the emissions to ensure sustainable

In 2024, Shukla, Singh and Kumar built the advanced GIS based vehicular emission inventory model which

includes exhaust and fugitive emissions. Their research illustrated a more precise spatial description of the

pollution sources and highlighted the importance of using a geospatial approach, complementing that by

Ghosal et al., (2025), developed a novel gridded emission inventory method that uses a super-resolution satellite-

This study emphasizes the study of two cities attached to the state of Maharashtra, one is Nagpur and the other

In most instances, the primary data sources of the present study are secondary data obtained from different

governmental agencies, transportation departments, environmental groups and public databases. The traffic data,

vehicle registration statistics, road network information, fuel consumption statistics and emission factors were

obtained from official reports and published literature. To investigate the effect of meteorological parameters on

vehicular emissions, the data pertaining to temperature, humidity, wind speed, precipitation, and weather

conditions for each season was taken from various meteorological databases including the India Meteorological

The study included various vehicle types including two-, three-, passenger cars, buses, light commercial, heavy-

duty vehicles and fuel types using petrol, diesel, CNG and electric. Temporal variations were also emphasised

by adding seasonal data. The data collected were pre-processed and validated and incorporated into the emission

inventory approaches along with advanced machine learning and deep learning approaches. A hybrid approach

comprising of a bottom-up and a top-down methodology was used for the estimation of vehicles emissions and

The results showed a significant difference in the numbers of registered vehicles - there are 5.99 million against

1.81 million for Mumbai and Nagpur respectively. In both cities, vehicles accounted for the majority of the

vehicle fleet, with passenger cars making up a significant portion of vehicle emissions, as did heavy vehicles.

Emissions of all pollutants (CO, NOx, PM₂.₅, PM₁₀, HC and CO₂) were significantly higher in Mumbai,

The seasonal analysis showed that seasonal vehicle emissions change over time, exhibiting highest emission at

winter season and lowest emission in monsoon season. The differences in these variations are caused by

variations in meteorological conditions, traffic and atmospheric dispersion. In addition, it showed that even

though they have a relatively small population, heavy-duty vehicles and passenger cars are high emitters of

The obtained spatial analysis with the grid technique was seen to have effectively identified emission hotspots

traffic image and video data with an accuracy score of 96.5%. Our findings show that using an AI-based method

yield more precise, efficient and scalable emission inventory development solutions than survey-based

In summary, the study highlights the power of integrating emission inventory approaches, GIS spatial analysis,

and AI techniques to develop a comprehensive framework for evaluating urban vehicular emissions. The built

model could be used in designing environmental monitoring, transportation planning and sustainable processes

for urban development. In addition, the results highlight the importance of developing cleaner automobiles and

application technology, enforcing stricter emission control measures, building a better public transport system,

and implementing smart traffic management methods to combat urban pollution originated from vehicle exhausts