AMAnju S. Mohan

Anju S. Mohan

*This is an algorithmically generated author record based on publicly available information.

Anju S. Mohan has a total of 3 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Sparse autoencoder-based ensemble model for particulate matter estimation using outdoor images

    by Anju S Mohan, Lizy Abraham, M Sabeena

    Air pollution poses a serious threat to public health and environmental sustainability. In India, the sparse distribution and high cost of conventional air quality monitoring stations limit fine-grained pollution assessment. This paper presents a sparse autoencoder-based ensemble regression framework for estimating PM 2.5 concentrations using outdoor images, augmented with meteorological and traffic data. An L1-regularized sparse convolutional autoencoder (SCAE) is employed to extract compact and regression-oriented image representations. These features are fused with auxiliary weather and traffic parameters and reduced using principal component analysis. A stacked ensemble regressor comprising support vector regression, k-nearest neighbors, and random forest models, with light gradient boosting machine as the meta-learner, is used for final PM 2.5 estimation. A new dataset, AirSetTvm, consisting of 8,488 annotated daytime outdoor images collected in Thiruvananthapuram, India, is introduced. Extensive experiments demonstrate that the proposed method outperforms classical CNN-based and hybrid deep learning baselines. Statistical significance analysis confirms that the performance improvements are robust. The proposed framework offers a cost-effective and scalable alternative to complement existing air quality monitoring infrastructure.

  • Open Access

    Sparse autoencoder-based ensemble model for particulate matter estimation using outdoor images

    by Anju S Mohan, Lizy Abraham

    Air pollution is a significant environmental threat faced by the world today. With each passing day, the air we breathe becomes increasingly contaminated, leading to severe health issues for individuals. Unfortunately, the existing air pollution monitoring stations are both expensive and insufficiently distributed. Consequently, the effectiveness of air pollution mitigation measures is hindered by inadequate data, underscoring the necessity for cost-effective alternatives. This paper introduces a sparse autoencoder-based ensemble model for estimating particulate matter concentrations using outdoor images. First, an L1 regularized sparse convolutional autoencoder compresses and extracts the pertinent features from images. Subsequently, these extracted image features are combined with weather and traffic data, followed by dimensionality reduction through principal component analysis. The final step involves a stacked ensemble of regression models to estimate PM2.5 concentrations. The ensemble incorporates support vector regression, k-nearest neighbor, and random forest regressor as base learners, with the light gradient boost machine acting as the meta-learner. An extensive dataset of 8488 single-scene outdoor images, named 'AirSetTvm,' has been meticulously collected and labeled with corresponding ground truth values derived from the continuous monitoring station. Encouraging results from this model, in comparison with other deep learning models in the existing literature, suggest that the proposed approach is a viable, cost-effective alternative for estimating PM2.5 concentrations.


  • Open Access

    An ensemble deep learning approach for air quality estimation in Delhi, India

    by Anju S Mohan, Lizy Abraham

    South Asian megacities are significant contributors to the degrading air quality. In highly populated northern India, Delhi is a major hotspot for air pollutants that influence health and climate. Effective mitigation of air pollution is impeded by inadequate estimation which emphasizes the need for cost-effective alternatives. This paper proposes an ensemble model based on transformer and Convolutional Neural Network (CNN) models to estimate air quality from images and weather parameters in Delhi. A Data Efficient Image transformer (DeiT) is fine-tuned with outdoor images, and parallelly dark-channel prior extracted from images are fed to a CNN model. Additionally, a 1-dimensional CNN is trained with meteorological features to improve accuracy. The predictions from these three parallel branches are then fused with ensemble learning to classify images into six Air Quality Index (AQI) classes and estimate the AQI value. To train and validate the proposed model, an image dataset is collected from Delhi, India termed ‘AirSetDelhi’ and properly labeled with ground-truth AQI values. Experiments conducted on the dataset demonstrate that the proposed model outperforms other deep learning networks in the literature. The model achieved an overall accuracy of 89.28% and a Cohen Kappa score of 0.856 for AQI classification, while it obtained an RMSE of 47.36 and an R2 value of 0.861 for AQI estimation, demonstrating efficacy in both tasks. As a regional estimation model based on images and weather features, the proposed model offers an alternative feasible approach for air quality estimation.

  • An Ensemble Deep Learning Model for Forecasting Hourly PM 2.5 Concentrations

    by Anju S. Mohan, Lizy Abraham

    Air pollution has become an environmental threat globally. Therefore, air pollution estimation and prediction are crucial for effecting timely control measures. This paper proposes an ensemble deep particulate forecaster (EDPF) to predict hourly PM2.5 concentrations for a day ahead. Air quality data from Thiruvananthapuram, Kerala (an Indian state), is used as a case study to demonstrate the model. The effect of other pollutant concentrations and various meteorological parameters on PM2.5 prediction is first studied using Pearson correlation analysis and recursive feature elimination method. Thus, the most relevant features for the forecast of PM2.5 concentrations are identified. Then preliminary PM2.5 predictions are made using two base predictors: an attention-based convolutional neural network - bidirectional long short-term memory (CNN-BiLSTM) model and a bidirectional long short-term memory model (BiLSTM). Finally, a feed-forward deep neural network (DNN) model combines the preliminary predictions yielding the final PM2.5 predictions. The proposed model is validated for long-term forecasting of PM2.5 at Thiruvananthapuram, and the model attains root mean squared error of 12.96 µg/m3 and mean absolute error of 9.28 µg/m3 for forecasting 24 h ahead PM2.5 concentrations. The experimental results demonstrate that the model performs better compared to various deep learning models in the literature. The effectiveness of the model in all four seasons is also evaluated, which reveals winter is the most challenging season to forecast.