M. Nagaraju
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
M. Nagaraju has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Priyanka Chawla
Department of Computer Science and Engineering, National Institute of Technology, Warangal, India
3 shared publications A Meghana
Institute of Aeronautical Engineering,Department of CSE(AIML),Hyderabad,Telangana,India
1 shared publicationAlsani Bhavika Reddy
1 shared publicationBojja Lavanya
1 shared publicationC.P. Sai Theja
Institute of Aeronautical Engineering,Department of CSE(AIML),Hyderabad,Telangana,India
1 shared publicationG Veera Sreeja
Institute of Aeronautical Engineering,Department of CSE(AIML),Hyderabad,Telangana,India
1 shared publication
Latest Publications
- Open Access
- 1 July 2026
- Published by Elsevier in SSRN Electronic Journal
Smart Digital Soil Mapping and Sustainable Land-Use Planning in Precision Farming
To ensure precision farming, nutrient optimization, and sustainable agricultural land-use planning, soil fertility assessment is necessary. Traditional soil testing available offers sound information but without constant space representation they would be ineffective in the identification of the nutrient deficient areas at regional level. This paper presents a explainable machine learning based digital soil mapping framework to assess the soils fertility at the district level using soil nutrient data, centroid coordinates based on GIS and supervised regression models. The data consisted of 875 records of the soil in 101 districts with four targeted soil properties; pH, nitrogen, phosphorus, and potassium. District centroid coordinates and district codes were used as spatial predictors to model nutrient variability. Four machine learning models (Random Forest, XGBoost, LightGBM and CatBoost) were evaluated through mean absolute error, root mean square error, coefficient of determination and cross-validation. LightGBM was best in terms of nitrogen prediction with an R2 of 0.8527, and Random Forest also achieved the best performance in the prediction of pH, phosphorus and potassium with R2 of 0.4452, 0.6472 and 0.6488 respectively. The concept of explainability was integrated using SHAP analysis and permutation feature importance. The projected soil characteristics were further transformed into the digital soil maps, classes of nutrient fertility, management zones and categories in sustainable land-use. The findings suggest that the proposed framework can aid the nutrition management of the districts and the agricultural planning of the regions.
- 1 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSmart Digital Soil Mapping and Sustainable Land-Use Planning in Precision Farming
To ensure precision farming, nutrient optimization, and sustainable agricultural land-use planning, soil fertility assessment is necessary. Traditional soil testing available offers sound information but without constant space representation they would be ineffective in the identification of the nutrient deficient areas at regional level. This paper presents a explainable machine learning based digital soil mapping framework to assess the soils fertility at the district level using soil nutrient data, centroid coordinates based on GIS and supervised regression models. The data consisted of 875 records of the soil in 101 districts with four targeted soil properties; pH, nitrogen, phosphorus, and potassium. District centroid coordinates and district codes were used as spatial predictors to model nutrient variability. Four machine learning models (Random Forest, XGBoost, LightGBM and CatBoost) were evaluated through mean absolute error, root mean square error, coefficient of determination and cross-validation. LightGBM was best in terms of nitrogen prediction with an R2 of 0.8527, and Random Forest also achieved the best performance in the prediction of pH, phosphorus and potassium with R2 of 0.4452, 0.6472 and 0.6488 respectively. The concept of explainability was integrated using SHAP analysis and permutation feature importance. The projected soil characteristics were further transformed into the digital soil maps, classes of nutrient fertility, management zones and categories in sustainable land-use. The findings suggest that the proposed framework can aid the nutrition management of the districts and the agricultural planning of the regions.
- 1 April 2026
- Published by Springer Nature
- p. 320-334
Self-learning-Based Decision Support System for Sustainable Crop Recommendation and Land Suitability Analysis
- 1 April 2026
- Published by Springer Nature
- p. 320-334
Self-learning-Based Decision Support System for Sustainable Crop Recommendation and Land Suitability Analysis
- 4 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1151-1158
Data-Driven Weather Forecasting and Climate Modeling Using Hybrid Modeling Techniques
Traditional meteorological models tend not to be updated to accommodate nonlinear and dynamic climate evolution, restricting their applicability in high-stakes areas. This paper introduces a hybrid, parallelized forecasting system that combines Auto Regressive Integrated Moving Average (ARIMA), Random Forest, and Long Short-term Memory (LSTM) models. Guided by logic gate-inspired pitch estimation algorithms, the model adapts adaptive selection with weighted variance penalties for unweighted model inclusion. A Self-Organizing Map (SOM) supports the system's computational efficiency through data dimensionality management. Measured on actual data, the model outperforms conventional and hybrid baselines consistently, providing better accuracy, generalization, and robustness to environmental variations. The new hybrid model outperformed in terms of accuracy, with Random Forest having the maximum $\mathrm{R}^{\mathrm{2}}$ value ($\text{0. 9 9 3 4}$) followed by LSTM (0.9292) and detecting seasonal trends but losing out in multivariate treatment. Even with such improvements, difficulties such as high computational expenses, real-time adaptation, and explainability continue. Real-time dynamic adaptation of the model, implementation of Explainable AI, and investigation into using quantum computing for scalable weather prediction solutions are topics of interest in future research.- 4 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1151-1158
Data-Driven Weather Forecasting and Climate Modeling Using Hybrid Modeling Techniques
Traditional meteorological models tend not to be updated to accommodate nonlinear and dynamic climate evolution, restricting their applicability in high-stakes areas. This paper introduces a hybrid, parallelized forecasting system that combines Auto Regressive Integrated Moving Average (ARIMA), Random Forest, and Long Short-term Memory (LSTM) models. Guided by logic gate-inspired pitch estimation algorithms, the model adapts adaptive selection with weighted variance penalties for unweighted model inclusion. A Self-Organizing Map (SOM) supports the system's computational efficiency through data dimensionality management. Measured on actual data, the model outperforms conventional and hybrid baselines consistently, providing better accuracy, generalization, and robustness to environmental variations. The new hybrid model outperformed in terms of accuracy, with Random Forest having the maximum $\mathrm{R}^{\mathrm{2}}$ value ($\text{0. 9 9 3 4}$) followed by LSTM (0.9292) and detecting seasonal trends but losing out in multivariate treatment. Even with such improvements, difficulties such as high computational expenses, real-time adaptation, and explainability continue. Real-time dynamic adaptation of the model, implementation of Explainable AI, and investigation into using quantum computing for scalable weather prediction solutions are topics of interest in future research.- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1367-1372
Transfer Learning Based Detection Model for COVID-19 Using Computed Tomography Scanned Images
COVID-19 pandemic has revealed the need for reliable and rapid diagnostic capabilities. An image based deep transfer learning framework is proposed in current research for the detection of COVID-19 from Computed tomography scan images. In this research work, the focus is on comparing three most commonly used architectures namely ResNet50, VGG16 and VGG19 to find out the best model for identifying COVID-19 cases. With the benefit of transfer learning, each model is specialized on a dataset of chest CT scans, which allows fast trainings with discovered high performance even if the data limitations. The study evaluates the models by their Accuracy, Precision, Recall and computational efficiency. Recently, the vgg19 model is checked with ResNet50 and VGG16 models for its competitive performance in experimental results.
- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1367-1372
Transfer Learning Based Detection Model for COVID-19 Using Computed Tomography Scanned Images
COVID-19 pandemic has revealed the need for reliable and rapid diagnostic capabilities. An image based deep transfer learning framework is proposed in current research for the detection of COVID-19 from Computed tomography scan images. In this research work, the focus is on comparing three most commonly used architectures namely ResNet50, VGG16 and VGG19 to find out the best model for identifying COVID-19 cases. With the benefit of transfer learning, each model is specialized on a dataset of chest CT scans, which allows fast trainings with discovered high performance even if the data limitations. The study evaluates the models by their Accuracy, Precision, Recall and computational efficiency. Recently, the vgg19 model is checked with ResNet50 and VGG16 models for its competitive performance in experimental results.
- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1849-1853
Performance Optimization of U-Net Network in Segmentation of Nerves Image Dataset
This research explores the performance and optimization of U-Net architectures for segmenting nerve images. U-Nets are prevalent in medical image analysis due to their effectiveness in capturing detailed features and spatial relationships. Our focus is on enhancing U-Net models by fine-tuning hyperparameters, applying data augmentation techniques, and utilizing advanced loss functions to boost segmentation accuracy and computational efficiency. The study compares various U-Net variants, evaluating their effectiveness in preserving the delicate structures of nerve tissues. The results indicate that optimized U-Nets deliver high accuracy with reduced processing times, making them ideal for real-time medical applications. This research provides valuable insights into the practical use of U-Nets for nerve image segmentation and suggests avenues for future improvements.
- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1849-1853
Performance Optimization of U-Net Network in Segmentation of Nerves Image Dataset
This research explores the performance and optimization of U-Net architectures for segmenting nerve images. U-Nets are prevalent in medical image analysis due to their effectiveness in capturing detailed features and spatial relationships. Our focus is on enhancing U-Net models by fine-tuning hyperparameters, applying data augmentation techniques, and utilizing advanced loss functions to boost segmentation accuracy and computational efficiency. The study compares various U-Net variants, evaluating their effectiveness in preserving the delicate structures of nerve tissues. The results indicate that optimized U-Nets deliver high accuracy with reduced processing times, making them ideal for real-time medical applications. This research provides valuable insights into the practical use of U-Nets for nerve image segmentation and suggests avenues for future improvements.
- 20 October 2024
- Published by Springer Nature
- p. 135-145
Deep Learning-Based Maize Crop Disease Detection and Remedial Recommendation System
- 20 October 2024
- Published by Springer Nature
- p. 135-145
Deep Learning-Based Maize Crop Disease Detection and Remedial Recommendation System
- 1 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on AgriFood Electronics
- Vol. 2 (2) , 627-637
Deep Learning-Based Maize Crop Disease Classification Model in Telangana Region of South India
One of India's main crops, maize, accounts for 2–3% of global production. Disease detection in maize fields has become increasingly difficult due to a lack of knowledge about disease symptoms. Furthermore, manual disease detection methods take a lot of time and are not effective. Recent developments in convolutional neural networks (CNNs) have exhibited remarkable performance in disease recognition and classification. A CNN is a deep learning technique that extracts the features from an image and performs the disease classification effectively. The optimization of hyperparameters is a tedious problem that impacts the performance of a model. The main purpose of the present research is to support future research to configure suitable hyperparameters to a model. In the present work, a deep CNN is proposed for the classification of seven different diseases of maize crop. Several hyperparameters, such as image size, batch size, number of epochs, optimizers, learning rate, kernel size, and number of hidden layers, were tested with various values in the experimental approach. The obtained results show that running the model for 200 epochs improved the classification accuracy with 87.44%. It also states that choosing input image sizes of 168 × 168 and 224 × 224 resulted in a good classification accuracy of 84.66% and 85.23%, respectively. The proposed deep CNN model has attained 85.83% classification accuracy with the Adam optimizer and a learning rate of 0.001. However, the results achieved by other optimizers, such as root-mean-square propagation (81.95%) and stochastic gradient descent (79.66%), are not better when compared with the Adam optimizer. Finally, the results have provided a better knowledge in selecting appropriate hyperparameters to the application of plant disease classification.
- 1 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on AgriFood Electronics
- Vol. 2 (2) , 627-637
Deep Learning-Based Maize Crop Disease Classification Model in Telangana Region of South India
One of India's main crops, maize, accounts for 2–3% of global production. Disease detection in maize fields has become increasingly difficult due to a lack of knowledge about disease symptoms. Furthermore, manual disease detection methods take a lot of time and are not effective. Recent developments in convolutional neural networks (CNNs) have exhibited remarkable performance in disease recognition and classification. A CNN is a deep learning technique that extracts the features from an image and performs the disease classification effectively. The optimization of hyperparameters is a tedious problem that impacts the performance of a model. The main purpose of the present research is to support future research to configure suitable hyperparameters to a model. In the present work, a deep CNN is proposed for the classification of seven different diseases of maize crop. Several hyperparameters, such as image size, batch size, number of epochs, optimizers, learning rate, kernel size, and number of hidden layers, were tested with various values in the experimental approach. The obtained results show that running the model for 200 epochs improved the classification accuracy with 87.44%. It also states that choosing input image sizes of 168 × 168 and 224 × 224 resulted in a good classification accuracy of 84.66% and 85.23%, respectively. The proposed deep CNN model has attained 85.83% classification accuracy with the Adam optimizer and a learning rate of 0.001. However, the results achieved by other optimizers, such as root-mean-square propagation (81.95%) and stochastic gradient descent (79.66%), are not better when compared with the Adam optimizer. Finally, the results have provided a better knowledge in selecting appropriate hyperparameters to the application of plant disease classification.