Ragavesh Dhandapani
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ragavesh Dhandapani has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
Gordon Morison
Glasgow Caledonian University,Department of Computing,Glasgow,UK,G40BA
2 shared publications- Gordon Morison
School of Computing, Engineering and Built Environment, Glasgow Caledonian University, 70 Cowcaddens Road, Glasgow G4 0BA, UK
2 shared publications Imene Mitiche
Glasgow Caledonian University,Department of Computing,Glasgow,UK,G40BA
2 shared publications- Imene Mitiche
Glasgow Caledonian University
2 shared publications - N. Aravind
Department of Civil and Environmental Engineering, National University of Science and Technology, Muscat, Sultanate of Oman
2 shared publications - Scott G. McMeekin
School of Engineering and Built Environment, Glasgow Caledonian University, Glasgow G4 0BA, UK
2 shared publications
Latest Publications
- Open Access
Preliminary Field Evaluation of a Low-Cost IoT Workflow for Dissolved Oxygen Monitoring and Short-Horizon Forecasting in Nile Tilapia Aquaculture
Short-term fluctuations in dissolved oxygen are difficult to capture in warm outdoor Nile tilapia (Oreochromis niloticus) ponds using periodic manual measurements, yet they strongly influence fish performance and farm management. This study presents a preliminary field evaluation of a low-cost IoT workflow for dissolved oxygen monitoring and short-horizon forecasting in pond-based tilapia culture. An ESP32-based sensing node continuously measured dissolved oxygen, temperature, and pH, transmitted readings to a cloud backend, and generated short-horizon forecasts from 5 min aggregated windows. During live validation from 1 to 10 April 2026, the 30 min forecast achieved a mean absolute error of 0.783 mg/L and directional accuracy of 60.23%, with only modest improvement over a persistence baseline. The 6 h forecast achieved 1.109 mg/L and 53.82%, respectively, indicating limited predictive value at the extended horizon. An extended 47-day field deployment (May–June 2026) captured four sensor-recorded low-DO events and two documented power outages, causing sensor downtime and providing additional field-deployment evidence. These results demonstrate the engineering feasibility of the integrated workflow, but they do not establish robust operational forecasting validity because the data were collected from one pond, high-frequency records were temporally correlated, and independent reference-meter validation was not available. The study is, therefore, best interpreted as a proof-of-concept field evaluation that identifies practical requirements for future low-cost aquaculture forecasting systems.
Open AccessPreliminary Field Evaluation of a Low-Cost IoT Workflow for Dissolved Oxygen Monitoring and Short-Horizon Forecasting in Nile Tilapia Aquaculture
Short-term fluctuations in dissolved oxygen are difficult to capture in warm outdoor Nile tilapia (Oreochromis niloticus) ponds using periodic manual measurements, yet they strongly influence fish performance and farm management. This study presents a preliminary field evaluation of a low-cost IoT workflow for dissolved oxygen monitoring and short-horizon forecasting in pond-based tilapia culture. An ESP32-based sensing node continuously measured dissolved oxygen, temperature, and pH, transmitted readings to a cloud backend, and generated short-horizon forecasts from 5 min aggregated windows. During live validation from 1 to 10 April 2026, the 30 min forecast achieved a mean absolute error of 0.783 mg/L and directional accuracy of 60.23%, with only modest improvement over a persistence baseline. The 6 h forecast achieved 1.109 mg/L and 53.82%, respectively, indicating limited predictive value at the extended horizon. An extended 47-day field deployment (May–June 2026) captured four sensor-recorded low-DO events and two documented power outages, causing sensor downtime and providing additional field-deployment evidence. These results demonstrate the engineering feasibility of the integrated workflow, but they do not establish robust operational forecasting validity because the data were collected from one pond, high-frequency records were temporally correlated, and independent reference-meter validation was not available. The study is, therefore, best interpreted as a proof-of-concept field evaluation that identifies practical requirements for future low-cost aquaculture forecasting systems.
- Open Access
- 1 June 2026
- Published by IOP Publishing in IOP Conference Series: Earth and Environmental Science
- Vol. 1640 (1) , 012001
A Deep Learning-Based System for Multi-Lane, Vehicle-Type Traffic Counting using Real-Time Video Analysis
This study investigates a deep learning-based multi-lane, vehicle-type traffic counting system as a decision-support tool for pollution and carbon emission management in congested urban corridors, where rising traffic volumes contribute significantly to local air pollution and transport-related greenhouse gas emissions. The proposed method employs a real-time video analysis framework that integrates a YOLO-based object detection model with object tracking techniques to detect, classify, and count vehicles across multiple lanes, generating lane-wise and class-wise traffic profiles that can be coupled with emission factors to support high-resolution emission estimation. The system is evaluated using offline video footage captured from a multi-lane roadway in Oman, with vehicle counts manually verified to assess performance. Experimental results show that the YOLOv10s-based model achieves a maximum accuracy of 98% over a 15-minute video segment, demonstrating reliable operation under realistic traffic conditions. By providing accurate, disaggregated traffic data, the framework can enhance pollution and carbon emission analysis, support the design of smart traffic management measures, and contribute to data-driven policies aimed at reducing vehicle emissions and improving urban air quality.
- 1 June 2026
- Published by IOP Publishing in IOP Conference Series: Earth and Environmental Science
- Vol. 1640 (1) , 012001
Open AccessA Deep Learning-Based System for Multi-Lane, Vehicle-Type Traffic Counting using Real-Time Video Analysis
This study investigates a deep learning-based multi-lane, vehicle-type traffic counting system as a decision-support tool for pollution and carbon emission management in congested urban corridors, where rising traffic volumes contribute significantly to local air pollution and transport-related greenhouse gas emissions. The proposed method employs a real-time video analysis framework that integrates a YOLO-based object detection model with object tracking techniques to detect, classify, and count vehicles across multiple lanes, generating lane-wise and class-wise traffic profiles that can be coupled with emission factors to support high-resolution emission estimation. The system is evaluated using offline video footage captured from a multi-lane roadway in Oman, with vehicle counts manually verified to assess performance. Experimental results show that the YOLOv10s-based model achieves a maximum accuracy of 98% over a 15-minute video segment, demonstrating reliable operation under realistic traffic conditions. By providing accurate, disaggregated traffic data, the framework can enhance pollution and carbon emission analysis, support the design of smart traffic management measures, and contribute to data-driven policies aimed at reducing vehicle emissions and improving urban air quality.
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Auto Road Inspector: Road Damage Recognition and Reporting System
A high-quality road network plays an important role in a country’s economic development. Well-maintained road network improves citizens’ living conditions and satisfaction. Hence, proper maintenance of the road network is highly essential. The proposed work in this article, ‘Auto Road Inspector’, is an innovative automated system designed to enhance road maintenance plans. The proposed system effectively detects road damage, such as ditches, potholes, and cracks, using Roboflow Object Detection v3.0, YOLO-NAS and a Convolutional Neural Network (CNN). The proposed work involves collecting high-quality images of various road conditions, followed by preprocessing and categorisation into different road defects. We used the following labels: 'Holes', 'Cracks', and 'Normal', which were labelled in the Roboflow Platform. We also developed a CNN and trained it on a custom dataset of images for damage detection and classification, supplemented by the YOLO-NAS model for its rapid processing and high accuracy. The findings reveal that our method is effective, achieving a mAP of 80.1%. Future work could expand the dataset across regions and different conditions and implement it in real time.
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Auto Road Inspector: Road Damage Recognition and Reporting System
A high-quality road network plays an important role in a country’s economic development. Well-maintained road network improves citizens’ living conditions and satisfaction. Hence, proper maintenance of the road network is highly essential. The proposed work in this article, ‘Auto Road Inspector’, is an innovative automated system designed to enhance road maintenance plans. The proposed system effectively detects road damage, such as ditches, potholes, and cracks, using Roboflow Object Detection v3.0, YOLO-NAS and a Convolutional Neural Network (CNN). The proposed work involves collecting high-quality images of various road conditions, followed by preprocessing and categorisation into different road defects. We used the following labels: 'Holes', 'Cracks', and 'Normal', which were labelled in the Roboflow Platform. We also developed a CNN and trained it on a custom dataset of images for damage detection and classification, supplemented by the YOLO-NAS model for its rapid processing and high accuracy. The findings reveal that our method is effective, achieving a mAP of 80.1%. Future work could expand the dataset across regions and different conditions and implement it in real time.
- 18 December 2025
- Published by Elsevier in Journal of the Indian Chemical Society
- Vol. 103 (1)
Eco-friendly extraction of nano alumina from aluminium foil: Characterization and antibacterial study
- 18 December 2025
- Published by Elsevier in Journal of the Indian Chemical Society
- Vol. 103 (1)
Eco-friendly extraction of nano alumina from aluminium foil: Characterization and antibacterial study
- Open Access
- 27 July 2025
- Published by Springer Nature in Discover Concrete and Cement
- Vol. 1 (1) , 1-25
Application of artificial intelligence in prediction of mechanical properties of sustainable plastic pavement blocks and validation through experimental analysiss
- 27 July 2025
- Published by Springer Nature in Discover Concrete and Cement
- Vol. 1 (1) , 1-25
Open AccessApplication of artificial intelligence in prediction of mechanical properties of sustainable plastic pavement blocks and validation through experimental analysiss
- Open Access
- 26 May 2025
- Published by Springer Nature in Cureus
- Vol. 17 (5) , e84816
Automated Workflow for Processing and Classifying Dental Radiographs: A Hands-On Approach
Background Classifying large dental radiographic datasets enables efficient data management and retrieval, facilitating quick access to specific types of radiographs for clinical or research purposes. It also supports advanced analytics, research, and the development of Artificial Intelligence (AI) tools. This study aimed to develop an automated workflow to improve the efficiency of dental radiograph classification. The workflow covers the entire process, from retrieving Digital Imaging and Communication in Medicine (DICOM) files to converting them into Joint Photographic Experts Group (JPEG) format and classifying them using Convolutional Neural Networks (CNNs) on a large dataset. Materials and methods This cross-sectional machine learning study was conducted using 48,329 dental radiographs to develop an automated classification workflow using CNNs. The workflow involved retrieving 48,329 DICOM files, standardizing them to a uniform size, and converting them to JPEG using the Pydicom library. Image preprocessing, including normalization, prepared the images for machine learning analysis. Various models, such as ResNet-50, AlexNet, and custom CNN models, were trained, validated, and tested on distinct datasets. Results These models were then deployed to classify a dataset of 48329 images. AlexNet demonstrated the highest performance, with a 95.98% detection rate and no errors, while ResNet-50 achieved 92.3% accuracy with 194 errors, and the custom CNN model showed a 77.25% detection rate with 1,623 errors. Conclusion The study established an effective automated workflow for dental radiograph classification, demonstrating that CNN models significantly improve classification accuracy and efficiency.
- 26 May 2025
- Published by Springer Nature in Cureus
- Vol. 17 (5) , e84816
Open AccessAutomated Workflow for Processing and Classifying Dental Radiographs: A Hands-On Approach
Background Classifying large dental radiographic datasets enables efficient data management and retrieval, facilitating quick access to specific types of radiographs for clinical or research purposes. It also supports advanced analytics, research, and the development of Artificial Intelligence (AI) tools. This study aimed to develop an automated workflow to improve the efficiency of dental radiograph classification. The workflow covers the entire process, from retrieving Digital Imaging and Communication in Medicine (DICOM) files to converting them into Joint Photographic Experts Group (JPEG) format and classifying them using Convolutional Neural Networks (CNNs) on a large dataset. Materials and methods This cross-sectional machine learning study was conducted using 48,329 dental radiographs to develop an automated classification workflow using CNNs. The workflow involved retrieving 48,329 DICOM files, standardizing them to a uniform size, and converting them to JPEG using the Pydicom library. Image preprocessing, including normalization, prepared the images for machine learning analysis. Various models, such as ResNet-50, AlexNet, and custom CNN models, were trained, validated, and tested on distinct datasets. Results These models were then deployed to classify a dataset of 48329 images. AlexNet demonstrated the highest performance, with a 95.98% detection rate and no errors, while ResNet-50 achieved 92.3% accuracy with 194 errors, and the custom CNN model showed a 77.25% detection rate with 1,623 errors. Conclusion The study established an effective automated workflow for dental radiograph classification, demonstrating that CNN models significantly improve classification accuracy and efficiency.
- 3 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Intelligent Road Safety System: Integrating Machine Learning for Enhanced Speed Regulation and License Plate Verification
Road safety remains a critical issue globally, with speeding identified as a leading cause of traffic accidents. Traditional speed monitoring systems often fail to capture violations occurring between checkpoints. This study introduces an Intelligent Road Safety System that integrates an Average Speed Camera (ASC) with License Plate Recognition (LPR) technology. The proposed system uses YOLOvll to detect and recognize license plates with high accuracy in real-time. The system calculates average speeds by time stamping vehicle movements at predefined checkpoints. The average speed is then compared with the speed limit set for the specific road to identify over speeding. We also created a master database with the plate number and registration details to validate the vehicle's registration status. The proposed system is tested under different conditions such as low light, glare, and obstructions. Experimental results demonstrate a detection accuracy of mAP@50: 99.3% and recognition accuracy of mAP@50-95: 99.4%, with inference time less than 40ms. This study contributes to minimize speeding-related accidents, supporting Oman Vision 2040's goals of safer roads and resilient infrastructure.
- 3 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Intelligent Road Safety System: Integrating Machine Learning for Enhanced Speed Regulation and License Plate Verification
Road safety remains a critical issue globally, with speeding identified as a leading cause of traffic accidents. Traditional speed monitoring systems often fail to capture violations occurring between checkpoints. This study introduces an Intelligent Road Safety System that integrates an Average Speed Camera (ASC) with License Plate Recognition (LPR) technology. The proposed system uses YOLOvll to detect and recognize license plates with high accuracy in real-time. The system calculates average speeds by time stamping vehicle movements at predefined checkpoints. The average speed is then compared with the speed limit set for the specific road to identify over speeding. We also created a master database with the plate number and registration details to validate the vehicle's registration status. The proposed system is tested under different conditions such as low light, glare, and obstructions. Experimental results demonstrate a detection accuracy of mAP@50: 99.3% and recognition accuracy of mAP@50-95: 99.4%, with inference time less than 40ms. This study contributes to minimize speeding-related accidents, supporting Oman Vision 2040's goals of safer roads and resilient infrastructure.
- Open Access
- 1 October 2024
- Published by IOP Publishing in IOP Conference Series: Earth and Environmental Science
- Vol. 1401 (1)
IoT-Enabled Street Dustbin Monitoring System Based on Machine Learning
In this article, a novel method is proposed by using the power of the Internet of Things (IoT) and Machine Learning (ML) to monitor and manage street dustbins effectively. The proposed innovative solution aimed at optimizing waste management processes and improper disposing of the waste in street dustbins. The system involves the installation of sensors in dustbins, which collect real-time data about the fill level of the bins. The capability of the proposed model allows for dynamic adjustment of waste collection schedules, reducing unnecessary trips and thus saving time, energy, and resources. In addition to that, the model is also trained in various waste disposal scenario such as improper disposal, massive garbage dumps, and e-waste. The proposed model is able to predict and report the trained conditions with the 100% accuracy. The system is implemented using Raspberry Pi 4, a single-board computer with the necessary hardware interface. A suitable web interface has been developed to oversee the status of dustbins and provide optimum travel route for collection of waste from the street dustbins. The proposed system not only enhances operational efficiency but also contributes to environmental sustainability.
- 1 October 2024
- Published by IOP Publishing in IOP Conference Series: Earth and Environmental Science
- Vol. 1401 (1)
Open AccessIoT-Enabled Street Dustbin Monitoring System Based on Machine Learning
In this article, a novel method is proposed by using the power of the Internet of Things (IoT) and Machine Learning (ML) to monitor and manage street dustbins effectively. The proposed innovative solution aimed at optimizing waste management processes and improper disposing of the waste in street dustbins. The system involves the installation of sensors in dustbins, which collect real-time data about the fill level of the bins. The capability of the proposed model allows for dynamic adjustment of waste collection schedules, reducing unnecessary trips and thus saving time, energy, and resources. In addition to that, the model is also trained in various waste disposal scenario such as improper disposal, massive garbage dumps, and e-waste. The proposed model is able to predict and report the trained conditions with the 100% accuracy. The system is implemented using Raspberry Pi 4, a single-board computer with the necessary hardware interface. A suitable web interface has been developed to oversee the status of dustbins and provide optimum travel route for collection of waste from the street dustbins. The proposed system not only enhances operational efficiency but also contributes to environmental sustainability.
- 11 September 2024
- Published by Glasgow Caledonian University
Application of Signal Processing Techniques in High Voltage Equipment Fault Detection Using Partial Discharge Signal
Partial Discharge (PD) detection in High Voltage (HV) equipment is an essential technique for assessing the insulation condition in the power system apparatus. The presence of PD is one of the critical indications for the defect and deterioration of the insulating material. Effective signal processing techniques are required to acquire, analyse, and classify PD signals into a specific fault type. The PD measurement device is prone to be affected by various noise sources, which leads to misinterpretation of the PD signal. This research focuses on developing a suitable algorithm for denoising PD signals and proposes a novel feature extraction process for classifying PD signals. The signal decomposition-based denoising algorithms in various engineering disciplines have become popular, and such techniques are being applied to PD signals. A novel PD signal denoising using adaptive decomposition, optimized by entropy and statistical parameters is proposed. Furthermore, a novel entropy-driven PD signal classification method is proposed. Initially, the PD signal is decomposed using various adaptive decomposition methods to get Intrinsic Mode Functions (IMFs). Then, IMFs are analysed using statistical or information theory measures such as Kurtosis, Mutual Information (MI) and Dispersion Entropy (DE). The IMFs are classified as noise, noise–dominant and signal IMFs, based on Kurtosis and MI values. The DE value of IMFs is used to select the regularisation parameter of Group-Sparse Total Variation (GSTV). The noise–dominant IMFs are subjected to the GSTV method, and the signal IMFs are added to the output vector to reconstruct the denoised output. The noisy IMFs are discarded for further processing. The proposed methods are applied under different input signal-noise ratios, lengths, and noise sources. The proposed denoising methods with the results and findings are presented and discussed, along with the performance metrics related to the filters. In addition, Tsallis entropy and a novel Generalized Gaussian Distribution – Refined Composite Multiscale Dispersion Entropy (GGD–RCMDE) based features are proposed for the time series signal. The suggested feature extraction methods are applied to the PD dataset to extract the features from various PD signals such as ‘corona’, ‘surface’ and ‘void’. A novel preprocessing method followed by the min-redundancy max-relevance (mRMR) feature selection method is applied to select sensitive features. A multiclass classifier model is trained using the most sensitive features to classify PD faults. The robust experimental analysis of the feature extraction method was carried out and reported. The performance metrics are computed using the results obtained through the experiments. The effectiveness of the proposed denoising and classification methods was presented with a suitable discussion and conclusion.
- 11 September 2024
- Published by Glasgow Caledonian University
Application of Signal Processing Techniques in High Voltage Equipment Fault Detection Using Partial Discharge Signal
Partial Discharge (PD) detection in High Voltage (HV) equipment is an essential technique for assessing the insulation condition in the power system apparatus. The presence of PD is one of the critical indications for the defect and deterioration of the insulating material. Effective signal processing techniques are required to acquire, analyse, and classify PD signals into a specific fault type. The PD measurement device is prone to be affected by various noise sources, which leads to misinterpretation of the PD signal. This research focuses on developing a suitable algorithm for denoising PD signals and proposes a novel feature extraction process for classifying PD signals. The signal decomposition-based denoising algorithms in various engineering disciplines have become popular, and such techniques are being applied to PD signals. A novel PD signal denoising using adaptive decomposition, optimized by entropy and statistical parameters is proposed. Furthermore, a novel entropy-driven PD signal classification method is proposed. Initially, the PD signal is decomposed using various adaptive decomposition methods to get Intrinsic Mode Functions (IMFs). Then, IMFs are analysed using statistical or information theory measures such as Kurtosis, Mutual Information (MI) and Dispersion Entropy (DE). The IMFs are classified as noise, noise–dominant and signal IMFs, based on Kurtosis and MI values. The DE value of IMFs is used to select the regularisation parameter of Group-Sparse Total Variation (GSTV). The noise–dominant IMFs are subjected to the GSTV method, and the signal IMFs are added to the output vector to reconstruct the denoised output. The noisy IMFs are discarded for further processing. The proposed methods are applied under different input signal-noise ratios, lengths, and noise sources. The proposed denoising methods with the results and findings are presented and discussed, along with the performance metrics related to the filters. In addition, Tsallis entropy and a novel Generalized Gaussian Distribution – Refined Composite Multiscale Dispersion Entropy (GGD–RCMDE) based features are proposed for the time series signal. The suggested feature extraction methods are applied to the PD dataset to extract the features from various PD signals such as ‘corona’, ‘surface’ and ‘void’. A novel preprocessing method followed by the min-redundancy max-relevance (mRMR) feature selection method is applied to select sensitive features. A multiclass classifier model is trained using the most sensitive features to classify PD faults. The robust experimental analysis of the feature extraction method was carried out and reported. The performance metrics are computed using the results obtained through the experiments. The effectiveness of the proposed denoising and classification methods was presented with a suitable discussion and conclusion.
- 8 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Belt Drive Condition Monitoring using Generalised Gaussian Distribution based Entropy Features
Belt drives are important and commonly used in industry to transmit power. In this paper, the vibration analysis is carried out on the centrifugal fan belt drive under different operating conditions. The real-time vibration signals were acquired for analysing and predicting the healthy and fault conditions at different operating speeds. Initially, the acquired signals were segmented, and the feature of each segment was calculated. An unique Generalized Gaussian Distribution - Refined Composite Multiscale Dispersion Entropy is considered for extracting the features with different shape parameters $\alpha$ . The primary feature is chosen from the feature set with the smallest standard deviation for a specific data set. Finally, the primary features of different vibration signals are then used to train a Neural Network to predict the healthy and faulty conditions of the belt drive. In comparison to the other approaches taken into consideration in this study, the suggested method has the best accuracy, precision, and recall.- 8 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Belt Drive Condition Monitoring using Generalised Gaussian Distribution based Entropy Features
Belt drives are important and commonly used in industry to transmit power. In this paper, the vibration analysis is carried out on the centrifugal fan belt drive under different operating conditions. The real-time vibration signals were acquired for analysing and predicting the healthy and fault conditions at different operating speeds. Initially, the acquired signals were segmented, and the feature of each segment was calculated. An unique Generalized Gaussian Distribution - Refined Composite Multiscale Dispersion Entropy is considered for extracting the features with different shape parameters $\alpha$ . The primary feature is chosen from the feature set with the smallest standard deviation for a specific data set. Finally, the primary features of different vibration signals are then used to train a Neural Network to predict the healthy and faulty conditions of the belt drive. In comparison to the other approaches taken into consideration in this study, the suggested method has the best accuracy, precision, and recall.