Pub Date : 2024-07-01DOI: 10.1109/TAFE.2024.3416221
Divyansh Thakur;Vikram Kumar
In this work, we developed and enhanced an artificial intelligence (AI)-centered hardware framework. This framework integrates the Nvidia Jetson Nano processing unit with a Depth AI camera. Our primary goal was to create an improved version of the YOLOv7 algorithm to quantify apple fruits using edge computing resources. We curated a dataset of 9,000 images of apple fruits to support this effort. Within the enhanced YOLOv7 architecture, we introduced a novel dual attention mechanism called the Global-SE Unified Attention Mechanism (GSEAM). This mechanism was designed to improve the accuracy of object detection by combining spatial and channel-oriented attention mechanisms, significantly enhancing the model.s contextual understanding and object recognition in various settings. The incorporation of GSEAM, along with the Gaussian Error Linear Unit activation function, was a deliberate effort to boost the YOLOv7 architecture.s ability to capture intricate contextual details and hierarchical features. Our system.s performance was rigorously evaluated across six key performance metrics and compared with other pretrained models. We achieved a precision of 99.54%, recall of 98.94%, F1-score of 99.71%, and average precision of 99.13%. This system has proven to be a valuable tool for real-time apple fruit counting, with practical applications for farmers.
{"title":"FruitVision: Dual-Attention Embedded AI System for Precise Apple Counting Using Edge Computing","authors":"Divyansh Thakur;Vikram Kumar","doi":"10.1109/TAFE.2024.3416221","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3416221","url":null,"abstract":"In this work, we developed and enhanced an artificial intelligence (AI)-centered hardware framework. This framework integrates the Nvidia Jetson Nano processing unit with a Depth AI camera. Our primary goal was to create an improved version of the YOLOv7 algorithm to quantify apple fruits using edge computing resources. We curated a dataset of 9,000 images of apple fruits to support this effort. Within the enhanced YOLOv7 architecture, we introduced a novel dual attention mechanism called the Global-SE Unified Attention Mechanism (GSEAM). This mechanism was designed to improve the accuracy of object detection by combining spatial and channel-oriented attention mechanisms, significantly enhancing the model.s contextual understanding and object recognition in various settings. The incorporation of GSEAM, along with the Gaussian Error Linear Unit activation function, was a deliberate effort to boost the YOLOv7 architecture.s ability to capture intricate contextual details and hierarchical features. Our system.s performance was rigorously evaluated across six key performance metrics and compared with other pretrained models. We achieved a precision of 99.54%, recall of 98.94%, F1-score of 99.71%, and average precision of 99.13%. This system has proven to be a valuable tool for real-time apple fruit counting, with practical applications for farmers.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"445-459"},"PeriodicalIF":0.0,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142408682","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-28DOI: 10.1109/TAFE.2024.3414953
Stephany Valarezo-Plaza;Julio Torres-Tello;Keshav D. Singh;Steve J. Shirtliffe;S. Deivalakshmi;Seok-Bum Ko
The escalating global demand for food, coupled with challenges in sustaining crop production, deteriorating ocean health, and depleting natural resources, underscores the critical role of agricultural technology. This article addresses the imperative of developing an optimal deep-learning model for predicting canola crop yield using hyperspectral images captured by drone flights. Our primary objective is to identify the most efficient model in terms of performance and size, considering the storage limitations on edge devices like Raspberry Pi 4 (RPi4). We start with the baseline 1D _ CNN model, which achieves an $R^{2}$ score of 0.82, and compress it into the proposed fs_model (fp32). To achieve the compression, we apply pruning through sparsity and feature selection using SHAP values. Further reduction in model size is accomplished by quantizing the weights of the proposed model to a lower precision, such as int16. This combined approach substantially decreases the proposed model's size by approximately 92.6% and inference time by approximately ×9013 in comparison to the baseline 1D _ CNN model. In addition, we propose the novel fsp_model posit(8,3) that uses posit quantization to further reduce the computation requirements compared to the proposed fs_model (int16). Our findings indicate that the utilization of posit numbers enables us to shrink the model size to 94% of the original base model, while only reducing the $R^{2}$ score by 5.7%.
{"title":"A Novel Optimized Deep Learning Model for Canola Crop Yield Prediction on Edge Devices","authors":"Stephany Valarezo-Plaza;Julio Torres-Tello;Keshav D. Singh;Steve J. Shirtliffe;S. Deivalakshmi;Seok-Bum Ko","doi":"10.1109/TAFE.2024.3414953","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3414953","url":null,"abstract":"The escalating global demand for food, coupled with challenges in sustaining crop production, deteriorating ocean health, and depleting natural resources, underscores the critical role of agricultural technology. This article addresses the imperative of developing an optimal deep-learning model for predicting canola crop yield using hyperspectral images captured by drone flights. Our primary objective is to identify the most efficient model in terms of performance and size, considering the storage limitations on edge devices like Raspberry Pi 4 (RPi4). We start with the baseline 1D\u0000<italic>_</i>\u0000CNN model, which achieves an \u0000<inline-formula><tex-math>$R^{2}$</tex-math></inline-formula>\u0000 score of 0.82, and compress it into the proposed \u0000<italic>fs_model</i>\u0000 (fp32). To achieve the compression, we apply pruning through sparsity and feature selection using SHAP values. Further reduction in model size is accomplished by quantizing the weights of the proposed model to a lower precision, such as int16. This combined approach substantially decreases the proposed model's size by approximately 92.6% and inference time by approximately ×9013 in comparison to the baseline 1D\u0000<italic>_</i>\u0000CNN model. In addition, we propose the novel \u0000<italic>fsp_model</i>\u0000 posit(8,3) that uses posit quantization to further reduce the computation requirements compared to the proposed \u0000<italic>fs_model</i>\u0000 (int16). Our findings indicate that the utilization of posit numbers enables us to shrink the model size to 94% of the original base model, while only reducing the \u0000<inline-formula><tex-math>$R^{2}$</tex-math></inline-formula>\u0000 score by 5.7%.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"436-444"},"PeriodicalIF":0.0,"publicationDate":"2024-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142408782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-26DOI: 10.1109/TAFE.2024.3408912
Jaime Giménez-Gallego;Jesús Martínez-del-Rincon;Pedro J. Blaya-Ros;Honorio Navarro-Hellín;Pedro J. Navarro;Roque Torres-Sánchez
Fruit harvest date prediction is crucial to optimize resource management, maximize quality, and minimize waste of this food. For this purpose, it is necessary to monitor the fruit ripening stage. However, current measurement procedures pose drawbacks for widespread field deployment: laboratory trials are manual, destructive and expensive; measurements with hand-held portable equipment in the field are very time consuming; and the use of remote sensing mobile platforms has a high operating cost. In this article, a low-cost autonomous fixed sensor for continuous on-tree monitoring of pomegranates is proposed. It is based on a computer vision system able to extract reliable fruit color and size estimations automatically. In addition, an empirical quantitative and qualitative study on the effectiveness of using image-based monitoring in comparison with in situ manual and lab-based measurements for pomegranates is provided in this work. Another contribution of this article is a harvest date prediction model that employs the fruit information collected from the images. Furthermore, a thorough quantitative evaluation of the proposed prediction model for the fruit harvest date was performed, being the median error of the best model of 3.5 days.
{"title":"Fruit Monitoring and Harvest Date Prediction Using On-Tree Automatic Image Tracking","authors":"Jaime Giménez-Gallego;Jesús Martínez-del-Rincon;Pedro J. Blaya-Ros;Honorio Navarro-Hellín;Pedro J. Navarro;Roque Torres-Sánchez","doi":"10.1109/TAFE.2024.3408912","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3408912","url":null,"abstract":"Fruit harvest date prediction is crucial to optimize resource management, maximize quality, and minimize waste of this food. For this purpose, it is necessary to monitor the fruit ripening stage. However, current measurement procedures pose drawbacks for widespread field deployment: laboratory trials are manual, destructive and expensive; measurements with hand-held portable equipment in the field are very time consuming; and the use of remote sensing mobile platforms has a high operating cost. In this article, a low-cost autonomous fixed sensor for continuous on-tree monitoring of pomegranates is proposed. It is based on a computer vision system able to extract reliable fruit color and size estimations automatically. In addition, an empirical quantitative and qualitative study on the effectiveness of using image-based monitoring in comparison with in situ manual and lab-based measurements for pomegranates is provided in this work. Another contribution of this article is a harvest date prediction model that employs the fruit information collected from the images. Furthermore, a thorough quantitative evaluation of the proposed prediction model for the fruit harvest date was performed, being the median error of the best model of 3.5 days.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"3 1","pages":"56-68"},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10572482","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143821875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-26DOI: 10.1109/TAFE.2024.3409166
Ankit Mittal;Ziyue Xu;Aatmesh Shrivastava
This article presents a robust, energy-efficient, and spectrum-aware infrastructure to support the Internet-of-Things (IoT) system deployed in precision agriculture, aiming to reduce power consumption to a level feasible for sustained operation using harvested energy alone. We present a system modeling-based approach to identify key optimizations, which are subsequently translated into a more feasible ultra-low power (ULP) IoT system implementation. Measurement results for ULP infrastructure components, including a ULP received signal strength detector and wake-up radio, implemented in a 65-nm CMOS technology, demonstrate power consumption in the range of a few nano-watts. In addition, we propose a lightweight energy-detection-based countermeasure against energy depletion attacks within IoT networks. We also suggest strategies for IoT sensor nodes to coexist within increasingly congested device networks while opportunistically enhancing their energy systems to potentially achieve self-powered IoT operation. Finally, we conduct a detailed analysis of power consumption in an IoT sensor deployed for sensing and monitoring, evaluating the feasibility of different energy systems, such as battery-based and energy harvesting solutions.
{"title":"Energy-Efficient, Secure, and Spectrum-Aware Ultra-Low Power Internet-of-Things System Infrastructure for Precision Agriculture","authors":"Ankit Mittal;Ziyue Xu;Aatmesh Shrivastava","doi":"10.1109/TAFE.2024.3409166","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3409166","url":null,"abstract":"This article presents a robust, energy-efficient, and spectrum-aware infrastructure to support the Internet-of-Things (IoT) system deployed in precision agriculture, aiming to reduce power consumption to a level feasible for sustained operation using harvested energy alone. We present a system modeling-based approach to identify key optimizations, which are subsequently translated into a more feasible ultra-low power (ULP) IoT system implementation. Measurement results for ULP infrastructure components, including a ULP received signal strength detector and wake-up radio, implemented in a 65-nm CMOS technology, demonstrate power consumption in the range of a few nano-watts. In addition, we propose a \u0000<italic>lightweight</i>\u0000 energy-detection-based countermeasure against energy depletion attacks within IoT networks. We also suggest strategies for IoT sensor nodes to coexist within increasingly congested device networks while opportunistically enhancing their energy systems to potentially achieve self-powered IoT operation. Finally, we conduct a detailed analysis of power consumption in an IoT sensor deployed for sensing and monitoring, evaluating the feasibility of different energy systems, such as battery-based and energy harvesting solutions.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"198-208"},"PeriodicalIF":0.0,"publicationDate":"2024-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430803","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Potatoes are one of the world's most popular and economically important crops. For many uses in agriculture, breeding, and trading, accurate recognition of potato breeds is important. In recent years, deep learning algorithms have become effective tools for breed recognition tasks using pictures, which inspires researchers to explore their potential for recognizing potato breeds. The paper presents extensive research on the application of deep learning for potato breed recognition. The recognition of potatoes has been effectively performed using the five state-of-the-art deep learning models VGG16, ResNet50, Mobile-Net, Inception-v3, and a customized CNN. These models have been modeled to differentiate between several potato breeds based on their unique visual characteristics, such as size, shape, color, texture, and skin pattern, by being trained on images of various potato breeds. The performance of each of the deep learning models is evaluated through thorough evaluation and testing. Among the models, the customized CNN model gives the best accuracy. The customized CNN model's accuracy is 94.84%. We do not just evaluate the accuracy but rather some other indicative metrics, such as F1-score, recall, and precision, too.
{"title":"Deep Learning Modeling for Potato Breed Recognition","authors":"Md. Ataur Rahman;Abbas Ali Khan;Md. Mehedi Hasan;Md. Sadekur Rahman;Md. Tarek Habib","doi":"10.1109/TAFE.2024.3406544","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3406544","url":null,"abstract":"Potatoes are one of the world's most popular and economically important crops. For many uses in agriculture, breeding, and trading, accurate recognition of potato breeds is important. In recent years, deep learning algorithms have become effective tools for breed recognition tasks using pictures, which inspires researchers to explore their potential for recognizing potato breeds. The paper presents extensive research on the application of deep learning for potato breed recognition. The recognition of potatoes has been effectively performed using the five state-of-the-art deep learning models VGG16, ResNet50, Mobile-Net, Inception-v3, and a customized CNN. These models have been modeled to differentiate between several potato breeds based on their unique visual characteristics, such as size, shape, color, texture, and skin pattern, by being trained on images of various potato breeds. The performance of each of the deep learning models is evaluated through thorough evaluation and testing. Among the models, the customized CNN model gives the best accuracy. The customized CNN model's accuracy is 94.84%. We do not just evaluate the accuracy but rather some other indicative metrics, such as F1-score, recall, and precision, too.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"419-427"},"PeriodicalIF":0.0,"publicationDate":"2024-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142408884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This article presents a comprehensive examination of the development of a non-contact measuring technique for determining fruit quality. Capacitance measurements were performed on soap (reference), banana, and nectarine samples across a frequency range of 5 Hz–200 kHz for banana and soap, and 10 Hz–1 MHz for nectarine. The data analysis revealed consistent trends in series capacitance ( $C_{s}$ ), indicating its suitability for future investigation. Additionally, temperature compensation improved data accuracy. Compensated capacitance data, obtained through linear fitting coefficients from the first 18 hours of data, showed distinct trends in banana samples, with a reduction of 6.76% on the first day and an additional 3.38% on the last day, illustrating the impact of aging. In contrast, the soap reference sample exhibited constant capacitance behavior over time. The response of the system to the presence and absence of the fruit sample and the effect of mass loss of the banana fruit on the Cs trends were also examined. The system's capacity to differentiate between undamaged and damaged samples was demonstrated after the investigation was expanded to include 51 nectarines. Following the impact damage, $C_{s}$ significantly increased, particularly one hour later, aligning with biochemical changes associated with mechanical damage. ANOVA, a type of multivariate analysis, highlighted the system's efficacy. The system demonstrated preserved damage detection even 24 hours after impact, despite temperature variations. This study provides valuable insights into non-contact measurement methods for potential industrial use, considering the effect of temperature and sample-specific analysis in the accurate evaluation of fruit quality.
{"title":"Capacitive Impedance Analysis for Noncontact Assessment of Fruit Quality and Ripening","authors":"Fahimeh Masoumi;Andrea Gottardo;Pietro Ibba;Matteo Caffini;Antonio Altana;Sundus Riaz;Luisa Petti;Paolo Lugli","doi":"10.1109/TAFE.2024.3406848","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3406848","url":null,"abstract":"This article presents a comprehensive examination of the development of a non-contact measuring technique for determining fruit quality. Capacitance measurements were performed on soap (reference), banana, and nectarine samples across a frequency range of 5 Hz–200 kHz for banana and soap, and 10 Hz–1 MHz for nectarine. The data analysis revealed consistent trends in series capacitance (\u0000<inline-formula><tex-math>$C_{s}$</tex-math></inline-formula>\u0000), indicating its suitability for future investigation. Additionally, temperature compensation improved data accuracy. Compensated capacitance data, obtained through linear fitting coefficients from the first 18 hours of data, showed distinct trends in banana samples, with a reduction of 6.76% on the first day and an additional 3.38% on the last day, illustrating the impact of aging. In contrast, the soap reference sample exhibited constant capacitance behavior over time. The response of the system to the presence and absence of the fruit sample and the effect of mass loss of the banana fruit on the Cs trends were also examined. The system's capacity to differentiate between undamaged and damaged samples was demonstrated after the investigation was expanded to include 51 nectarines. Following the impact damage, \u0000<inline-formula><tex-math>$C_{s}$</tex-math></inline-formula>\u0000 significantly increased, particularly one hour later, aligning with biochemical changes associated with mechanical damage. ANOVA, a type of multivariate analysis, highlighted the system's efficacy. The system demonstrated preserved damage detection even 24 hours after impact, despite temperature variations. This study provides valuable insights into non-contact measurement methods for potential industrial use, considering the effect of temperature and sample-specific analysis in the accurate evaluation of fruit quality.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"428-435"},"PeriodicalIF":0.0,"publicationDate":"2024-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10569992","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142408759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-19DOI: 10.1109/TAFE.2024.3402710
B. H. M. Imdaad;S. I. Jayalath;P. C. G. Mahiepala;M. K. T. Sampath;S. R. Munasinghe
This research presents an efficient fruit monitoring and orchard management system to replace the existing paper-based manual process. The new method is based on five state-of-the-art technologies, namely Radio Frequency IDentification (RFID), Wi-Fi network, Mobile App, Coud database/server, and Web Application. This paper presents the proper integration of these technologies to provide an effective, worker-friendly, and cost-effective solution to the problem. The proposed method starts by attaching RFID tags to each tree and each fruit and registering them in the cloud database. The cloud server visualizes the status of the orchard and implements inventory management. Workers use a hand device for tasks such as bagging, spraying, and plucking. The orchard manager carries out task assignments and worker deployment on the web application. Each worker gets notified of the assigned tasks on his hand device, and when such tasks are accomplished, the status is updated in the cloud database. Using this system, each fruit is monitored from the initial covering state to the final plucking state. On the contrary, the existing paper-based manual process suffers from improper spraying that leads to disease-spreading, infestations, and loss of yield. Due to the existing inefficiencies the price of fruit has gone up to a limit that is not affordable to the public, hence unprofitable to the grower as well. In this context, the proposed solution will help monitor and manage fruit orchards efficiently, which will increase the quality and quantity of the yield while lowering the cost of production.
这项研究提出了一种高效的水果监测和果园管理系统，以取代现有的纸质人工流程。新方法基于五项最先进的技术，即射频识别（RFID）、Wi-Fi 网络、移动应用程序、Coud 数据库/服务器和网络应用程序。本文介绍了这些技术的适当整合，以提供一个有效、方便工人且经济高效的解决方案。建议的方法首先是在每棵树和每个果实上安装 RFID 标签，并将其注册到云数据库中。云服务器将果园的状态可视化，并实施库存管理。工人使用手动设备完成套袋、喷洒和采摘等任务。果园管理员在网络应用程序上进行任务分配和工人部署。每个工人都会收到其手持设备上被分配任务的通知，任务完成后，状态会在云数据库中更新。使用该系统，每个水果从最初的覆盖状态到最后的采摘状态都会受到监控。相反，现有的基于纸张的人工流程存在喷洒不当的问题，导致病菌传播、虫害和产量损失。由于现有的低效率，水果价格上涨到了公众难以承受的程度，因此种植者也无利可图。在这种情况下，拟议的解决方案将有助于有效地监测和管理果园，从而提高产量的质量和数量，同时降低生产成本。
{"title":"RFID Based Fruit Monitoring and Orchard Management System","authors":"B. H. M. Imdaad;S. I. Jayalath;P. C. G. Mahiepala;M. K. T. Sampath;S. R. Munasinghe","doi":"10.1109/TAFE.2024.3402710","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3402710","url":null,"abstract":"This research presents an efficient fruit monitoring and orchard management system to replace the existing paper-based manual process. The new method is based on five state-of-the-art technologies, namely Radio Frequency IDentification (RFID), Wi-Fi network, Mobile App, Coud database/server, and Web Application. This paper presents the proper integration of these technologies to provide an effective, worker-friendly, and cost-effective solution to the problem. The proposed method starts by attaching RFID tags to each tree and each fruit and registering them in the cloud database. The cloud server visualizes the status of the orchard and implements inventory management. Workers use a hand device for tasks such as bagging, spraying, and plucking. The orchard manager carries out task assignments and worker deployment on the web application. Each worker gets notified of the assigned tasks on his hand device, and when such tasks are accomplished, the status is updated in the cloud database. Using this system, each fruit is monitored from the initial covering state to the final plucking state. On the contrary, the existing paper-based manual process suffers from improper spraying that leads to disease-spreading, infestations, and loss of yield. Due to the existing inefficiencies the price of fruit has gone up to a limit that is not affordable to the public, hence unprofitable to the grower as well. In this context, the proposed solution will help monitor and manage fruit orchards efficiently, which will increase the quality and quantity of the yield while lowering the cost of production.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"413-418"},"PeriodicalIF":0.0,"publicationDate":"2024-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142408880","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this article, an innovative deep learning-driven framework, adapted for the identification of diseases in strawberry plants, is proposed. Our approach encompasses a comprehensive embedded electronic system, incorporating sensor data acquisition and image capturing from the plants. These images are seamlessly transmitted to the cloud through a dedicated gateway for subsequent analysis. The research introduces a novel model, ResNet9-SE, a modified ResNet architecture featuring two squeeze-and-excitation (SE) blocks strategically positioned within the network to enhance performance. The key advantage gained is achieving fewer parameters and occupying less memory while preserving a high diagnosis accuracy. The proposed model is evaluated using in-house collected data and a publicly available dataset. The experimental outcomes demonstrate the exceptional classification accuracy of the ResNet9-SE model (99.7%), coupled with significantly reduced computation costs, affirming its suitability for deployment in embedded systems.
{"title":"Strawberry Disease Detection Through an Advanced Squeeze-and-Excitation Deep Learning Model","authors":"Jiayi Wu;Vahid Abolghasemi;Mohammad Hossein Anisi;Usman Dar;Andrey Ivanov;Chris Newenham","doi":"10.1109/TAFE.2024.3412285","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3412285","url":null,"abstract":"In this article, an innovative deep learning-driven framework, adapted for the identification of diseases in strawberry plants, is proposed. Our approach encompasses a comprehensive embedded electronic system, incorporating sensor data acquisition and image capturing from the plants. These images are seamlessly transmitted to the cloud through a dedicated gateway for subsequent analysis. The research introduces a novel model, ResNet9-SE, a modified ResNet architecture featuring two squeeze-and-excitation (SE) blocks strategically positioned within the network to enhance performance. The key advantage gained is achieving fewer parameters and occupying less memory while preserving a high diagnosis accuracy. The proposed model is evaluated using in-house collected data and a publicly available dataset. The experimental outcomes demonstrate the exceptional classification accuracy of the ResNet9-SE model (99.7%), coupled with significantly reduced computation costs, affirming its suitability for deployment in embedded systems.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"259-267"},"PeriodicalIF":0.0,"publicationDate":"2024-06-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430872","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-17DOI: 10.1109/TAFE.2024.3411269
Antonio Altana;Saleh Hamed;Paolo Lugli;Luisa Petti;Pietro Ibba
Insufficient availability of essential nutrients, such as iron, can impede plant growth, decrease crop productivity, and even lead to plant death. This is why it is crucial to employ proximal monitoring techniques to detect early signs of nutrient stress and prevent yield loss. In this study, we continuously monitored the stem impedance of eight tomato plants every hour for 38 days. This was done to observe the effects of iron stress by comparing these plants with those not under stress. The normalized impedance magnitude at 10 kHz reveals a noticeable divergence in the trend of impedance magnitude shortly after the removal of iron from the nutrient solution, clearly indicating the effect of iron stress on plant bioimpedance. Additionally, the Cole equivalent circuit model was employed to evaluate the electrical parameters of the impedance spectra. The fitting results exhibit an average root-mean-square error of 466.3 $Omega$ . Statistical analysis of the extracted circuit parameters shows significant differences between iron-stressed and control plants. Based on this hypothesis, the extracted circuit components have been used to train the machine learning classification model with several algorithms, to demonstrate that the multilayer perceptron is the best performing model, yielding 98% accuracy and 91% and 89% precision in identifying early and late stress, respectively. This research demonstrates the effectiveness of bioimpedance measurements in tracking iron stress in plants. Our findings highlight the usefulness of impedance measurements for monitoring iron stress in plants and provide insights into the physiological responses of tomato plants to nutrient deprivation by observing changes in bioimpedance circuit parameters over time.
{"title":"Monitoring Iron Stress in Tomato Plants Through Bioimpedance and Machine-Learning-Enhanced Classification Based on Circuit Component Analysis","authors":"Antonio Altana;Saleh Hamed;Paolo Lugli;Luisa Petti;Pietro Ibba","doi":"10.1109/TAFE.2024.3411269","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3411269","url":null,"abstract":"Insufficient availability of essential nutrients, such as iron, can impede plant growth, decrease crop productivity, and even lead to plant death. This is why it is crucial to employ proximal monitoring techniques to detect early signs of nutrient stress and prevent yield loss. In this study, we continuously monitored the stem impedance of eight tomato plants every hour for 38 days. This was done to observe the effects of iron stress by comparing these plants with those not under stress. The normalized impedance magnitude at 10 kHz reveals a noticeable divergence in the trend of impedance magnitude shortly after the removal of iron from the nutrient solution, clearly indicating the effect of iron stress on plant bioimpedance. Additionally, the Cole equivalent circuit model was employed to evaluate the electrical parameters of the impedance spectra. The fitting results exhibit an average root-mean-square error of 466.3 \u0000<inline-formula><tex-math>$Omega$</tex-math></inline-formula>\u0000. Statistical analysis of the extracted circuit parameters shows significant differences between iron-stressed and control plants. Based on this hypothesis, the extracted circuit components have been used to train the machine learning classification model with several algorithms, to demonstrate that the multilayer perceptron is the best performing model, yielding 98% accuracy and 91% and 89% precision in identifying early and late stress, respectively. This research demonstrates the effectiveness of bioimpedance measurements in tracking iron stress in plants. Our findings highlight the usefulness of impedance measurements for monitoring iron stress in plants and provide insights into the physiological responses of tomato plants to nutrient deprivation by observing changes in bioimpedance circuit parameters over time.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"190-197"},"PeriodicalIF":0.0,"publicationDate":"2024-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10559754","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-06-14DOI: 10.1109/TAFE.2024.3406648
Luca Barbisan;Giovanna Turvani;Fabrizio Riente
Honeybees play a pivotal role in maintaining global ecosystems and agricultural productivity through their indispensable contribution to crop pollination. However, the alarming rise in honeybee mortality, attributed to various stress factors including climate change, has highlighted the urgency of implementing effective monitoring strategies. Remote sensing of beehives emerges as a promising solution, with a focus on understanding and mitigating the impacts of these stressors. Differently from other approaches proposed in the literature, this study specifically explores the potential of lightweight machine learning models and the extraction of compressed feature to enable future deployment on microcontroller devices. The experimentation involves the application of support vector machines and neural network classifiers, considering the influence of variable audio chunk durations, the utilization of different hyperparameters and combining the audio recorded in several hives and available in different datasets.
{"title":"A Machine Learning Approach for Queen Bee Detection Through Remote Audio Sensing to Safeguard Honeybee Colonies","authors":"Luca Barbisan;Giovanna Turvani;Fabrizio Riente","doi":"10.1109/TAFE.2024.3406648","DOIUrl":"https://doi.org/10.1109/TAFE.2024.3406648","url":null,"abstract":"Honeybees play a pivotal role in maintaining global ecosystems and agricultural productivity through their indispensable contribution to crop pollination. However, the alarming rise in honeybee mortality, attributed to various stress factors including climate change, has highlighted the urgency of implementing effective monitoring strategies. Remote sensing of beehives emerges as a promising solution, with a focus on understanding and mitigating the impacts of these stressors. Differently from other approaches proposed in the literature, this study specifically explores the potential of lightweight machine learning models and the extraction of compressed feature to enable future deployment on microcontroller devices. The experimentation involves the application of support vector machines and neural network classifiers, considering the influence of variable audio chunk durations, the utilization of different hyperparameters and combining the audio recorded in several hives and available in different datasets.","PeriodicalId":100637,"journal":{"name":"IEEE Transactions on AgriFood Electronics","volume":"2 2","pages":"236-243"},"PeriodicalIF":0.0,"publicationDate":"2024-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10557729","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142430790","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：