Jayakrishnan Anandakrishnan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jayakrishnan Anandakrishnan has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Arun Kumar Sangaiah
National Yunlin University of Science and Technology
5 shared publications - Gui-Bin Bian
Chinese Academy of Sciences
3 shared publications - Mohammed J. F. Alenazi
College of Computer and Information Sciences, Department of Computer Engineering King Saud University Riyadh Saudi Arabia
3 shared publications - Salman Ali AlQahtani
New Emerging Technologies and 5G Network and Beyond Research Chair, Department of Computer Engineering, College of Computer and Information Sciences, King Saud University, Riyadh 11451, Saudi Arabia
3 shared publications Alkha Mohan
Indian Institute of Information Technology (IIIT) Kottayam
2 shared publications- Alkha Mohan
International Graduate School of Artificial Intelligence, National Yunlin University of Science and Technology, Yunlin, Taiwan
2 shared publications
Latest Publications
- 6 November 2026
- Published by Elsevier in Software Impacts
- Vol. 30
saga-activation: A Python package for Spatially Adaptive Gated Activation in convolutional neural networks
- 6 November 2026
- Published by Elsevier in Software Impacts
- Vol. 30
saga-activation: A Python package for Spatially Adaptive Gated Activation in convolutional neural networks
- 3 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (3) , 4033-4046
E-OptEEG: A Hybrid Ensemble Metaheuristic Feature Optimization for EEG-Based Sentiment Analysis on Resource-Constrained Edge Devices
Electroencephalogram (EEG) signal processing is essential for achieving accurate and efficient real-time edge computing, particularly in resource-constrained smart wearables. They demand lightweight and optimized solutions for rapid analysis and decision-making. However, the higher dimensionality of EEG signals introduces challenges of latency and computational cost on edge devices. This article presents E-OptEEG, a hybrid ensemble metaheuristic framework designed to bring edge-optimized EEG processing and feature selection specifically for low-powered devices. The E-OptEEG framework integrates the strengths of multiple evolutionary feature selection algorithms, leveraging an ensemble threshold voting scheme to combine their outputs and identify the most relevant features. Further, E-OptEEG employs a fruit-fly optimization algorithm with deep metric feature transformation based on cosine similarity to transform the refined feature set to a minimal latent space. E-OptEEG demonstrates its ability to identify and select the most relevant features, enabling effective emotion and sentiment analysis from EEG signals captured through edge-powered wearable devices. Experimental evaluations against state-of-the-art feature selection techniques on three different EEG-based behavior modeling datasets highlight the framework’s effectiveness, achieving an average accuracy exceeding 98% with an average accuracy improvement of 2.55%. The E-OptEEG framework exemplifies the potential for lightweight artificial intelligence solutions to enable real-time, resource-efficient decision-making in wearable electronics.
- 3 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (3) , 4033-4046
E-OptEEG: A Hybrid Ensemble Metaheuristic Feature Optimization for EEG-Based Sentiment Analysis on Resource-Constrained Edge Devices
Electroencephalogram (EEG) signal processing is essential for achieving accurate and efficient real-time edge computing, particularly in resource-constrained smart wearables. They demand lightweight and optimized solutions for rapid analysis and decision-making. However, the higher dimensionality of EEG signals introduces challenges of latency and computational cost on edge devices. This article presents E-OptEEG, a hybrid ensemble metaheuristic framework designed to bring edge-optimized EEG processing and feature selection specifically for low-powered devices. The E-OptEEG framework integrates the strengths of multiple evolutionary feature selection algorithms, leveraging an ensemble threshold voting scheme to combine their outputs and identify the most relevant features. Further, E-OptEEG employs a fruit-fly optimization algorithm with deep metric feature transformation based on cosine similarity to transform the refined feature set to a minimal latent space. E-OptEEG demonstrates its ability to identify and select the most relevant features, enabling effective emotion and sentiment analysis from EEG signals captured through edge-powered wearable devices. Experimental evaluations against state-of-the-art feature selection techniques on three different EEG-based behavior modeling datasets highlight the framework’s effectiveness, achieving an average accuracy exceeding 98% with an average accuracy improvement of 2.55%. The E-OptEEG framework exemplifies the potential for lightweight artificial intelligence solutions to enable real-time, resource-efficient decision-making in wearable electronics.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 157066-157080
Multimodal Sentinel-1/2 Data Fusion With a Dual-Path ViT-Kolmogorov–Arnold Network Pipeline for AI-Driven Flood Event Classification
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 157066-157080
Open AccessMultimodal Sentinel-1/2 Data Fusion With a Dual-Path ViT-Kolmogorov–Arnold Network Pipeline for AI-Driven Flood Event Classification
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18060-18075
HydroFedNet: An Intent-Based Unified Federated Framework for Multisource Water Quality Monitoring
Ensuring clean water availability is critical for sustainability and health. Conventional water quality assessments are limited by manual sampling, poor temporal resolution, and centralized data processing. This study proposes HydroFedNet, a multisource water quality monitoring framework that uses federated learning (FL) to integrate diverse data sources, including Landsat satellite imagery, RGB pond images, and Internet of Things (IoT) sensor streams. The spatiospectral transfer learning network (Spatiospectral TLNet), the color transfer learning network (Color TLNet), and the sensor convolutional neural network–temporal convolutional network (Sensor CNN–TCN) are fundamental models for HydroFedNet. Spatiospectral TLNet and Color TLNet leverage EfficientNetB3 for optimized, low-cost training, while Sensor CNN–TCN exploits improved temporal modeling. Models are trained locally and share weight updates with a central server, which builds a global model using the chosen FL strategy. FL strategies such as federated averaging (FedAvg), FL with temporally aware aggregation, and federated optimization (FedOpt) are evaluated with six objectives, including energy efficiency, fault tolerance, and handling of nonindependent and identically distributed (non-IID) data. FedLTA surpasses the 90% accuracy across all three models with less communication overhead, whereas FedOpt effectively handles non-IID data. HydroFedNet allows an optimal selection of an intent-aware FL strategy, allowing robust, scalable, and efficient water quality monitoring across heterogeneous environments.
- 31 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (9) , 18060-18075
HydroFedNet: An Intent-Based Unified Federated Framework for Multisource Water Quality Monitoring
Ensuring clean water availability is critical for sustainability and health. Conventional water quality assessments are limited by manual sampling, poor temporal resolution, and centralized data processing. This study proposes HydroFedNet, a multisource water quality monitoring framework that uses federated learning (FL) to integrate diverse data sources, including Landsat satellite imagery, RGB pond images, and Internet of Things (IoT) sensor streams. The spatiospectral transfer learning network (Spatiospectral TLNet), the color transfer learning network (Color TLNet), and the sensor convolutional neural network–temporal convolutional network (Sensor CNN–TCN) are fundamental models for HydroFedNet. Spatiospectral TLNet and Color TLNet leverage EfficientNetB3 for optimized, low-cost training, while Sensor CNN–TCN exploits improved temporal modeling. Models are trained locally and share weight updates with a central server, which builds a global model using the chosen FL strategy. FL strategies such as federated averaging (FedAvg), FL with temporally aware aggregation, and federated optimization (FedOpt) are evaluated with six objectives, including energy efficiency, fault tolerance, and handling of nonindependent and identically distributed (non-IID) data. FedLTA surpasses the 90% accuracy across all three models with less communication overhead, whereas FedOpt effectively handles non-IID data. HydroFedNet allows an optimal selection of an intent-aware FL strategy, allowing robust, scalable, and efficient water quality monitoring across heterogeneous environments.
- 4 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (5) , 7801-7814
Point-KAN: Leveraging Trustworthy AI for Reliable 3-D Point Cloud Completion With Kolmogorov-Arnold Networks for 6G-IoT Applications
3-D point clouds are data points defining the morphology of environments, and completion refers to the reconstruction of missing points. 6G Internet of Things (6G-IoT) connected with 3-D mapping devices needs reliable, consistent, high-fidelity real-time point cloud completion for accurate environment registration. Trustworthy AI, modeled with dependable deep learning (DL), enables reliable and robust point completion with spatial-geometrical consistency for deployment with 6G-IoT devices. Although several DL-based completion techniques are integrated with 6G-IoT devices, they have reliability issues, limiting key trustworthy AI characteristics. This research focuses on the reliability and robustness aspects of trustworthy AI to propose Point-Kolmogorov-Arnold Network (KAN), a dependable real-time 3-D point cloud completion model for 6G IoT-connected 3-D mapping devices. Point-KAN integrates multihead attention and KAN within the modules of attention enhanced-embedded feature collector (AEFC) and KAN-enhanced feature mapper (KEFM) for trustworthy point cloud completion. Empirical evaluations on the ShapeNet demonstrate the superiority of Point-KAN against state-of-the-art (SOTA). Results concrete Point-KAN’s evolution as a trustworthy AI framework ensures reliability and robustness for real-time deployment in 6G-IoT-connected devices, facilitating 3-D environment mapping.
- 4 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (5) , 7801-7814
Point-KAN: Leveraging Trustworthy AI for Reliable 3-D Point Cloud Completion With Kolmogorov-Arnold Networks for 6G-IoT Applications
3-D point clouds are data points defining the morphology of environments, and completion refers to the reconstruction of missing points. 6G Internet of Things (6G-IoT) connected with 3-D mapping devices needs reliable, consistent, high-fidelity real-time point cloud completion for accurate environment registration. Trustworthy AI, modeled with dependable deep learning (DL), enables reliable and robust point completion with spatial-geometrical consistency for deployment with 6G-IoT devices. Although several DL-based completion techniques are integrated with 6G-IoT devices, they have reliability issues, limiting key trustworthy AI characteristics. This research focuses on the reliability and robustness aspects of trustworthy AI to propose Point-Kolmogorov-Arnold Network (KAN), a dependable real-time 3-D point cloud completion model for 6G IoT-connected 3-D mapping devices. Point-KAN integrates multihead attention and KAN within the modules of attention enhanced-embedded feature collector (AEFC) and KAN-enhanced feature mapper (KEFM) for trustworthy point cloud completion. Empirical evaluations on the ShapeNet demonstrate the superiority of Point-KAN against state-of-the-art (SOTA). Results concrete Point-KAN’s evolution as a trustworthy AI framework ensures reliability and robustness for real-time deployment in 6G-IoT-connected devices, facilitating 3-D environment mapping.
- 30 April 2025
- Published by IGI Global Scientific Publishing
- p. 1-36
A Parallel Attention-Guided Generative Adversarial Network for Efficient Thin Cloud Removal From Satellite Imagery
Earth observation depends on the spatial-temporal data from satellites. Optical observations are often affected by random thin clouds. This cloud interference impacts the usefulness of satellite-based remote sensing in several application areas. This paper introduces a cloud-free reconstruction architecture based on a Generative Adversarial Network (GAN) that leverages spatial-attention mechanisms. The proposed Parallel Attention Guided Generative Adversarial Network for Efficient Thin Cloud Removal (PACR-GAN) integrates the benefits of the Convolutional Block Attention Module (CBAM) and the Coordinate Attention Module (CAM). When tested against the RICE-1 dataset, the proposed model demonstrated superior performance in terms of popular evaluation metrics when compared to existing methods. The model effectively reconstructed cloud-free images by focusing on critical features and spatial details, showing resilience to thin clouds.
- 30 April 2025
- Published by IGI Global Scientific Publishing
- p. 1-36
A Parallel Attention-Guided Generative Adversarial Network for Efficient Thin Cloud Removal From Satellite Imagery
Earth observation depends on the spatial-temporal data from satellites. Optical observations are often affected by random thin clouds. This cloud interference impacts the usefulness of satellite-based remote sensing in several application areas. This paper introduces a cloud-free reconstruction architecture based on a Generative Adversarial Network (GAN) that leverages spatial-attention mechanisms. The proposed Parallel Attention Guided Generative Adversarial Network for Efficient Thin Cloud Removal (PACR-GAN) integrates the benefits of the Convolutional Block Attention Module (CBAM) and the Coordinate Attention Module (CAM). When tested against the RICE-1 dataset, the proposed model demonstrated superior performance in terms of popular evaluation metrics when compared to existing methods. The model effectively reconstructed cloud-free images by focusing on critical features and spatial details, showing resilience to thin clouds.
- Open Access
- 31 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 6, 2837-2852
LCUT-Sv9: UAV-Assisted Powerline Inspection Framework with Secure Time-Sensitive Communication for Industry 5.0
Integrating Time-Sensitive Networking (TSN) in industrial wireless networks ensures reliability in data transmission. Automated powerline inspection using Unmanned Aerial Vehicles (UAVs) is an industrial application that requires time-constrained secure data exchanges with its ground station. However it is currently facing several critical challenges include handling massive data generated from onboard sensor, the limited copmutational capabilities, and ensuring secure communication over open wireless channels. Aiming to emphasize human-centric automation for enhanced operational safety and in alignment with the principles of Industry 5.0, this paper introduces a time-sensitive communication and detection framework for UAV-Industrial applications, named Latent-Coded UAV-IoT Transmission Segmentation YOLOv9 (LCUT-Sv9). It employs a modified lightweight Tiny-YOLOv9 model for powerline detection and utilizes autoencoder-based Latent-Coded UAV-IoT Transmission (LCUT) with the Message Queuing Telemetry Transport (MQTT) protocol for time-sensitive communication. The LCUT-Sv9 framework was evaluated using the widely adopted Transmission Towers / Power Lines Aerial-image (TTPLA) dataset, and demonstrated a significant detection accuracy improvement of 24.15% over state-of-the-art (SOTA) techniques. LCUT also ensures minimal data transmission, reducing encoded data by approximately 50%, making it compatible with the strict requirements of TSN-enabled industrial wireless networks.
- 31 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Communications Society
- Vol. 6, 2837-2852
Open AccessLCUT-Sv9: UAV-Assisted Powerline Inspection Framework with Secure Time-Sensitive Communication for Industry 5.0
Integrating Time-Sensitive Networking (TSN) in industrial wireless networks ensures reliability in data transmission. Automated powerline inspection using Unmanned Aerial Vehicles (UAVs) is an industrial application that requires time-constrained secure data exchanges with its ground station. However it is currently facing several critical challenges include handling massive data generated from onboard sensor, the limited copmutational capabilities, and ensuring secure communication over open wireless channels. Aiming to emphasize human-centric automation for enhanced operational safety and in alignment with the principles of Industry 5.0, this paper introduces a time-sensitive communication and detection framework for UAV-Industrial applications, named Latent-Coded UAV-IoT Transmission Segmentation YOLOv9 (LCUT-Sv9). It employs a modified lightweight Tiny-YOLOv9 model for powerline detection and utilizes autoencoder-based Latent-Coded UAV-IoT Transmission (LCUT) with the Message Queuing Telemetry Transport (MQTT) protocol for time-sensitive communication. The LCUT-Sv9 framework was evaluated using the widely adopted Transmission Towers / Power Lines Aerial-image (TTPLA) dataset, and demonstrated a significant detection accuracy improvement of 24.15% over state-of-the-art (SOTA) techniques. LCUT also ensures minimal data transmission, reducing encoded data by approximately 50%, making it compatible with the strict requirements of TSN-enabled industrial wireless networks.
- Open Access
- 2 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 1817-1826
STA-AgriNet: A Spatio-Temporal Attention Framework for Crop Type Mapping from Fused Multi-Sensor Multi-Temporal SITS
Precise and timely crop type mapping delivers insights into crop growth statistics and ensures food security for growing economies. Automated mapping is crucial in several agricultural applications, including crop wear assessment and yield forecasting. The high-resolution multispectral optical data can deliver essential spatial-spectral characteristics; however, these are typically impeded by unfavourable weather and obstructions, resulting in poor classification. Recent advancements in multi-sensor data fusion techniques have focused on integrating optical data with auxiliary Synthetic Aperture Radar (SAR) data to mitigate misclassification issues. However, current optical-SAR fusion techniques have yet to effectively address the incorporation of spatial-spectral characteristics with long-term temporal dependencies of Satellite Image Time Series (SITS). This paper proposes an optical-SAR deep fusion framework, STA-AgriNet, that integrates a U-Net encoder-decoder with spatial-temporal attention frameworks to enable superior extraction of long-term spatial-temporal dependencies for reliable crop-type mapping. The Spectral Spatial Feature Mapper (SSFM), Mixed Parallel Spatial Attention (MPSA), and Spatio-Temporal Attention Mapper (STAM) modules of the STA-AgriNet extract key classification-defining, discriminative patterns for semantic segmentation. The STA-AgriNet framework is evaluated against current state-of-the-art (SOTA) methods and demonstrates superior performance across multiple metrics, achieving an accuracy of 83.61% with a minimal inference time of 10.09 seconds and a compact parameter count of 0.97 million. The model also excels in other key evaluation metrics, establishing its overall effectiveness and efficiency compared to existing techniques.
- 2 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 1817-1826
Open AccessSTA-AgriNet: A Spatio-Temporal Attention Framework for Crop Type Mapping from Fused Multi-Sensor Multi-Temporal SITS
Precise and timely crop type mapping delivers insights into crop growth statistics and ensures food security for growing economies. Automated mapping is crucial in several agricultural applications, including crop wear assessment and yield forecasting. The high-resolution multispectral optical data can deliver essential spatial-spectral characteristics; however, these are typically impeded by unfavourable weather and obstructions, resulting in poor classification. Recent advancements in multi-sensor data fusion techniques have focused on integrating optical data with auxiliary Synthetic Aperture Radar (SAR) data to mitigate misclassification issues. However, current optical-SAR fusion techniques have yet to effectively address the incorporation of spatial-spectral characteristics with long-term temporal dependencies of Satellite Image Time Series (SITS). This paper proposes an optical-SAR deep fusion framework, STA-AgriNet, that integrates a U-Net encoder-decoder with spatial-temporal attention frameworks to enable superior extraction of long-term spatial-temporal dependencies for reliable crop-type mapping. The Spectral Spatial Feature Mapper (SSFM), Mixed Parallel Spatial Attention (MPSA), and Spatio-Temporal Attention Mapper (STAM) modules of the STA-AgriNet extract key classification-defining, discriminative patterns for semantic segmentation. The STA-AgriNet framework is evaluated against current state-of-the-art (SOTA) methods and demonstrates superior performance across multiple metrics, achieving an accuracy of 83.61% with a minimal inference time of 10.09 seconds and a compact parameter count of 0.97 million. The model also excels in other key evaluation metrics, establishing its overall effectiveness and efficiency compared to existing techniques.
- Open Access
- 25 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 2226-2238
Precise Spatial Prediction of Rice Seedlings From Large-Scale Airborne Remote Sensing Data Using Optimized Li-YOLOv9
Rice is pivotal in global food security and Rice seeding detection in precision agriculture is essential for optimizing crop productivity and efficient resource use. Currently, the spatial distribution and detection of rice seeding are manually done, which is time-consuming. As technology advances, the rapid progress in precision agriculture calls for innovative solutions like lightweight edge-computing vision models for unmanned aerial vehicle (UAV) intelligence. However, the limited computational capabilities of UAVs make it challenging to deploy complex object detection models onboard. This article proposes Li-YOLOv9, an efficient, lightweight, and precise object detection model for detection of rice seedlings from large-scale UAV remote sensing data. The proposed Li-YOLOv9 includes a 3-D feature adaptation module, convolutional block attention modules, coordinate attention module, and deepwise point convolution modules, which offer significant advantages over the original YOLOv9. The compact Li-YOLOv9 with approximately 9 million parameters is significantly lighter than the original YOLOv9 with 60 million parameters, making it ideal for resource-efficient onboard intelligence. Li-YOLOv9 is tested and validated against a real UAV rice seedling dataset. It showcases exceptional detection capabilities, achieving impressive metrics across various performance indicators, including a mean average precision of 99.60%, an $F1$ -score of 96.95%, an average recall of 96.70%, and an average precision of 97.20% in comparison to the state-of-the-art. Li-YOLOv9 is highly efficient and ideal for UAV deployment for real-time detection of rice seedlings from large-scale remote sensing data.
- 25 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 2226-2238
Open AccessPrecise Spatial Prediction of Rice Seedlings From Large-Scale Airborne Remote Sensing Data Using Optimized Li-YOLOv9
Rice is pivotal in global food security and Rice seeding detection in precision agriculture is essential for optimizing crop productivity and efficient resource use. Currently, the spatial distribution and detection of rice seeding are manually done, which is time-consuming. As technology advances, the rapid progress in precision agriculture calls for innovative solutions like lightweight edge-computing vision models for unmanned aerial vehicle (UAV) intelligence. However, the limited computational capabilities of UAVs make it challenging to deploy complex object detection models onboard. This article proposes Li-YOLOv9, an efficient, lightweight, and precise object detection model for detection of rice seedlings from large-scale UAV remote sensing data. The proposed Li-YOLOv9 includes a 3-D feature adaptation module, convolutional block attention modules, coordinate attention module, and deepwise point convolution modules, which offer significant advantages over the original YOLOv9. The compact Li-YOLOv9 with approximately 9 million parameters is significantly lighter than the original YOLOv9 with 60 million parameters, making it ideal for resource-efficient onboard intelligence. Li-YOLOv9 is tested and validated against a real UAV rice seedling dataset. It showcases exceptional detection capabilities, achieving impressive metrics across various performance indicators, including a mean average precision of 99.60%, an $F1$ -score of 96.95%, an average recall of 96.70%, and an average precision of 97.20% in comparison to the state-of-the-art. Li-YOLOv9 is highly efficient and ideal for UAV deployment for real-time detection of rice seedlings from large-scale remote sensing data.
- 29 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 11 (2) , 1197-1209
R-UAV-Net: Enhanced YOLOv4 With Graph-Semantic Compression for Transformative UAV Sensing in Paddy Agronomy
Common leaf diseases pose severe problems to the agricultural industry, particularly for paddy rice, a staple crop consumed worldwide, making early detection and rapid prevention crucial for maintaining both quality and yield. This research dwells on the object detection farmwork for identifying and localising paddy leaf diseases. Future-tech Unmanned Aerial Vehicles (UAVs) offer benefits such as reduced deployment costs, increased availability, enhanced operability, and improved geographical and temporal resolution. You Only Look Once (YOLO) models excel in disease part detection but require excessive computing. A severe challenge of UAV sensing is the resource-efficient collection, transmission and disease detection from this high-resolution ground data. This research addresses these issues by introducing a Graph-inspired encoder-decoder Semantic Compression (G-SC) coupled with enhanced YOLOv4 architecture for disease detection in paddy agronomy. The proposed R-UAV-Net is an improved YOLOv4 architecture incorporating various spatial and channel feature extraction blocks with attention mechanisms for revolutionizing precision farming. R-UAV-Net outperformed state-of-the-art (SOTA) techniques, showing a 0.69% improvement in mean average precision (mAP) and a 0.12 increase in F1 score over the best-performing leaf detection model.
- 29 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 11 (2) , 1197-1209
R-UAV-Net: Enhanced YOLOv4 With Graph-Semantic Compression for Transformative UAV Sensing in Paddy Agronomy
Common leaf diseases pose severe problems to the agricultural industry, particularly for paddy rice, a staple crop consumed worldwide, making early detection and rapid prevention crucial for maintaining both quality and yield. This research dwells on the object detection farmwork for identifying and localising paddy leaf diseases. Future-tech Unmanned Aerial Vehicles (UAVs) offer benefits such as reduced deployment costs, increased availability, enhanced operability, and improved geographical and temporal resolution. You Only Look Once (YOLO) models excel in disease part detection but require excessive computing. A severe challenge of UAV sensing is the resource-efficient collection, transmission and disease detection from this high-resolution ground data. This research addresses these issues by introducing a Graph-inspired encoder-decoder Semantic Compression (G-SC) coupled with enhanced YOLOv4 architecture for disease detection in paddy agronomy. The proposed R-UAV-Net is an improved YOLOv4 architecture incorporating various spatial and channel feature extraction blocks with attention mechanisms for revolutionizing precision farming. R-UAV-Net outperformed state-of-the-art (SOTA) techniques, showing a 0.69% improvement in mean average precision (mAP) and a 0.12 increase in F1 score over the best-performing leaf detection model.