Or Bar-Shira1, Yosef Cohen, T. Shoshan, A. Bechar, A. Sadowsky, Yuval Cohen, S. Berman
Highlights Medjool date fruit bunches can be modeled in 3D based on structural decomposition and the use of Bezier curves. The 3D model can be used for generating artificial image datasets of Medjool fruit bunches. The annotated image datasets can be used to develop robust algorithms for robotic Medjool date thinning. Algorithms for determining the required thinning length are a prerequisite for Medjool date thinning automation. Abstract. Medjool is a premium date cultivar, and the market demands high-quality fruits, for which specific horticultural practices, including timely and efficient fruitlet thinning, are required. Currently, thinning the fruitlets is one of the most labor-intensive tasks in the Medjool cultivation cycle, and there is a need to develop methods for automating the thinning process. An algorithm determining the required thinning is a prerequisite for advancing toward thinning automation. An annotated Medjool fruit bunch image dataset is necessary for developing such an algorithm using state-of-the-art machine learning methods. Acquiring such a dataset is difficult and costly. The difficulty can be alleviated by using synthetic images. However, current methods for generating synthetic plant images are unsuitable for Medjool dates due to their geometrical features. The current work suggests a method for generating artificial images of Medjool fruit bunches from a 3D model based on structural decomposition into plant parts and the use of Bezier curves. Nineteen model variables and their distributions were defined for fruit bunch model generation. The models and synthetic images generated based on the models were verified by two plant physiologists who are experts in Medjool date cultivation. Fruit-bunch features were extracted from the generated images and used for learning the required remaining length of the spikelets after thinning using kernel estimation. The estimation was tested for images of two whorl-period combinations (Top-Early and Middle-Middle). The average scaled absolute estimation errors for both periods were very low (less than 1%).
{"title":"Artificial Medjool Date Fruit Bunch Image Synthesis: Towards Thinning Automation","authors":"Or Bar-Shira1, Yosef Cohen, T. Shoshan, A. Bechar, A. Sadowsky, Yuval Cohen, S. Berman","doi":"10.13031/ja.15217","DOIUrl":"https://doi.org/10.13031/ja.15217","url":null,"abstract":"Highlights Medjool date fruit bunches can be modeled in 3D based on structural decomposition and the use of Bezier curves. The 3D model can be used for generating artificial image datasets of Medjool fruit bunches. The annotated image datasets can be used to develop robust algorithms for robotic Medjool date thinning. Algorithms for determining the required thinning length are a prerequisite for Medjool date thinning automation. Abstract. Medjool is a premium date cultivar, and the market demands high-quality fruits, for which specific horticultural practices, including timely and efficient fruitlet thinning, are required. Currently, thinning the fruitlets is one of the most labor-intensive tasks in the Medjool cultivation cycle, and there is a need to develop methods for automating the thinning process. An algorithm determining the required thinning is a prerequisite for advancing toward thinning automation. An annotated Medjool fruit bunch image dataset is necessary for developing such an algorithm using state-of-the-art machine learning methods. Acquiring such a dataset is difficult and costly. The difficulty can be alleviated by using synthetic images. However, current methods for generating synthetic plant images are unsuitable for Medjool dates due to their geometrical features. The current work suggests a method for generating artificial images of Medjool fruit bunches from a 3D model based on structural decomposition into plant parts and the use of Bezier curves. Nineteen model variables and their distributions were defined for fruit bunch model generation. The models and synthetic images generated based on the models were verified by two plant physiologists who are experts in Medjool date cultivation. Fruit-bunch features were extracted from the generated images and used for learning the required remaining length of the spikelets after thinning using kernel estimation. The estimation was tested for images of two whorl-period combinations (Top-Early and Middle-Middle). The average scaled absolute estimation errors for both periods were very low (less than 1%).","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"175 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74159053","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
J. Qin, Jeehwa Hong, Hyunjeong Cho, J. V. Van Kessel, I. Baek, K. Chao, M. Kim
Highlights A multimodal optical sensing system was developed for food safety applications. The prototype system can conduct dual-band Raman spectroscopy at 785 and 1064 nm. The system can automatically measure samples in Petri dishes or well plates. The system with AI software is promising for identifying species of foodborne bacteria. Abstract. A novel multimodal optical sensing system was developed for automated and intelligent food safety inspection. The system uses two pairs of compact point lasers and dispersive spectrometers at 785 and 1064 nm to realize dual-band Raman spectroscopy and imaging, which is suitable to measure samples generating low- and high-fluorescence interference signals, respectively. Automated spectral acquisition can be performed using a direct-drive XY moving stage for solid, powder, and liquid samples placed in customized well plates or randomly scattered in standard Petri dishes (e.g., bacterial colonies). Three LED lights (white backlight, UV ring light, and white ring light) and two miniature color cameras are used for machine vision measurements of samples in the Petri dishes using different combinations of illuminations and imaging modalities (e.g., transmission, fluorescence, and color). Real-time image processing and motion control techniques are used to implement automated sample counting, positioning, sampling, and synchronization functions. System software was developed using LabVIEW with integrated artificial intelligence functions able to identify and label interesting targets instantly. The system capability was demonstrated by an example application for rapid identification of five common foodborne bacteria, including Bacillus cereus, E. coli, Listeria monocytogenes, Staphylococcus aureus, and Salmonella spp.. Using a machine learning model based on a linear support vector machine, a classification accuracy of 98.6% was achieved using Raman spectra automatically collected from 222 bacterial colonies of the five species grown on nutrient nonselective agar in 90 mm Petri dishes. The entire system was built on a 30×45 cm2 breadboard, enabling it compact and portable and its use for field and on-site biological and chemical food safety inspection in regulatory and industrial applications. Keywords: Artificial intelligence, Automated sampling, Bacteria, Food safety, Machine learning, Machine vision, Raman, Sensing.
{"title":"A Multimodal Optical Sensing System for Automated and Intelligent Food Safety Inspection","authors":"J. Qin, Jeehwa Hong, Hyunjeong Cho, J. V. Van Kessel, I. Baek, K. Chao, M. Kim","doi":"10.13031/ja.15526","DOIUrl":"https://doi.org/10.13031/ja.15526","url":null,"abstract":"Highlights A multimodal optical sensing system was developed for food safety applications. The prototype system can conduct dual-band Raman spectroscopy at 785 and 1064 nm. The system can automatically measure samples in Petri dishes or well plates. The system with AI software is promising for identifying species of foodborne bacteria. Abstract. A novel multimodal optical sensing system was developed for automated and intelligent food safety inspection. The system uses two pairs of compact point lasers and dispersive spectrometers at 785 and 1064 nm to realize dual-band Raman spectroscopy and imaging, which is suitable to measure samples generating low- and high-fluorescence interference signals, respectively. Automated spectral acquisition can be performed using a direct-drive XY moving stage for solid, powder, and liquid samples placed in customized well plates or randomly scattered in standard Petri dishes (e.g., bacterial colonies). Three LED lights (white backlight, UV ring light, and white ring light) and two miniature color cameras are used for machine vision measurements of samples in the Petri dishes using different combinations of illuminations and imaging modalities (e.g., transmission, fluorescence, and color). Real-time image processing and motion control techniques are used to implement automated sample counting, positioning, sampling, and synchronization functions. System software was developed using LabVIEW with integrated artificial intelligence functions able to identify and label interesting targets instantly. The system capability was demonstrated by an example application for rapid identification of five common foodborne bacteria, including Bacillus cereus, E. coli, Listeria monocytogenes, Staphylococcus aureus, and Salmonella spp.. Using a machine learning model based on a linear support vector machine, a classification accuracy of 98.6% was achieved using Raman spectra automatically collected from 222 bacterial colonies of the five species grown on nutrient nonselective agar in 90 mm Petri dishes. The entire system was built on a 30×45 cm2 breadboard, enabling it compact and portable and its use for field and on-site biological and chemical food safety inspection in regulatory and industrial applications. Keywords: Artificial intelligence, Automated sampling, Bacteria, Food safety, Machine learning, Machine vision, Raman, Sensing.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74394392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Y. Xiong, Guoming Li, B. Ramirez, R. Burns, R. Gates
Highlights Draft EPA emission models for laying hen facilities were systematically evaluated. The models performed poorly on predicting the air pollutants when input variables were out of the NAEMS data range. A key finding was the unanticipated sensitivity of the draft model outputs to bird inventory and climate zones. Further revision and improvement may be necessary for draft models before they can be adopted by the egg industry. Abstract. In August 2021, the U.S. Environmental Protection Agency (EPA) released draft models to estimate daily NH3, H2S, PM10, PM2.5, and TSP emissions from egg-layer houses (high-rise and manure-belt) and manure storage using inputs of daily mean ambient temperature, relative humidity (RH), and hen inventory. These models were developed from refined datasets generated by the National Air Emissions Monitoring Study fieldwork completed in 2009. Notably, they do not include data for cage-free housing. Currently, 66% of U.S. laying hens are housed in cages; thus, these models, if adopted, will have a substantial impact on the U.S. egg industry. This study evaluated the EPA draft models’ robustness and assessed model outputs for egg production systems under differing climate scenarios. The EPA draft models distort emission factors for bird inventories to be lower or higher than those used to develop the models. With inventory held constant, the marginal influence of ambient temperature and RH on daily emissions varied substantially, with some values falling below the measurement detection threshold while others exceeding literature findings. For twelve representative U.S. locations representing differing climates, substantial differences in emission factors were found for bird inventories outside the range in the database. Annual emissions estimated from inventories used to develop the EPA models also varied by location. We conclude that the current draft EPA emission models cannot be used to the degree of precision that is suitable to apply to a wide range of layer facilities, particularly cage-free systems. Revisions are suggested to accommodate a greater range of climates, laying hen facility types, and inventories for practical emission estimations. Keywords: Air quality, Ammonia, Egg production, Emission model, Hydrogen sulfide, Particulate matter, Poultry.
{"title":"Evaluating Draft EPA Emissions Models for Laying Hen Facilities","authors":"Y. Xiong, Guoming Li, B. Ramirez, R. Burns, R. Gates","doi":"10.13031/ja.15237","DOIUrl":"https://doi.org/10.13031/ja.15237","url":null,"abstract":"Highlights Draft EPA emission models for laying hen facilities were systematically evaluated. The models performed poorly on predicting the air pollutants when input variables were out of the NAEMS data range. A key finding was the unanticipated sensitivity of the draft model outputs to bird inventory and climate zones. Further revision and improvement may be necessary for draft models before they can be adopted by the egg industry. Abstract. In August 2021, the U.S. Environmental Protection Agency (EPA) released draft models to estimate daily NH3, H2S, PM10, PM2.5, and TSP emissions from egg-layer houses (high-rise and manure-belt) and manure storage using inputs of daily mean ambient temperature, relative humidity (RH), and hen inventory. These models were developed from refined datasets generated by the National Air Emissions Monitoring Study fieldwork completed in 2009. Notably, they do not include data for cage-free housing. Currently, 66% of U.S. laying hens are housed in cages; thus, these models, if adopted, will have a substantial impact on the U.S. egg industry. This study evaluated the EPA draft models’ robustness and assessed model outputs for egg production systems under differing climate scenarios. The EPA draft models distort emission factors for bird inventories to be lower or higher than those used to develop the models. With inventory held constant, the marginal influence of ambient temperature and RH on daily emissions varied substantially, with some values falling below the measurement detection threshold while others exceeding literature findings. For twelve representative U.S. locations representing differing climates, substantial differences in emission factors were found for bird inventories outside the range in the database. Annual emissions estimated from inventories used to develop the EPA models also varied by location. We conclude that the current draft EPA emission models cannot be used to the degree of precision that is suitable to apply to a wide range of layer facilities, particularly cage-free systems. Revisions are suggested to accommodate a greater range of climates, laying hen facility types, and inventories for practical emission estimations. Keywords: Air quality, Ammonia, Egg production, Emission model, Hydrogen sulfide, Particulate matter, Poultry.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"3 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72830023","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Method for Zoning Corn Based on the NDVI and the Improved SOM-K-Means Algorithm","authors":"Xiaodong Di, X. Wang","doi":"10.13031/ja.15081","DOIUrl":"https://doi.org/10.13031/ja.15081","url":null,"abstract":"","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"281 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77495858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Estimating WEPP Cropland Erodibility Values From Soil Properties","authors":"W. Elliot, D. Flanagan","doi":"10.13031/ja.15218","DOIUrl":"https://doi.org/10.13031/ja.15218","url":null,"abstract":"","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77674925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Toby A. Adjuik, S. Nokes, M. Montross, M. Sama, O. Wendroth
Highlights In this study, six machine learning (ML) models were developed using a large database of soils to predict saturated hydraulic conductivity of these soils using easily measured soil characteristics. Tree-based regression models outperformed all other ML models tested. Neural networks were not suitable for predicting saturated hydraulic conductivity. Clay content, followed by bulk density, explained the highest amount of variation in the data of the predictors examined. Abstract. One of the most important soil hydraulic properties for modeling water transport in the vadose zone is saturated hydraulic conductivity. However, it is challenging to measure it in the field. Pedotransfer Functions (PTFs) are mathematical models that can predict saturated hydraulic conductivity (Ks) from easily measured soil characteristics. Though the development of PTFs for predicting Ks is not new, the tools and methods used to predict Ks are continuously evolving. Model performance depends on choosing soil features that explain the largest amount of Ks variance with the fewest input variables. In addition, the lack of interpretability in most “black box” machine learning models makes it difficult to extract practical knowledge as the machine learning process obfuscates the relationship between inputs and outputs in the PTF models. The objective of this study was to develop a set of new PTFs for predicting Ks using machine learning algorithms and a large database of over 8000 soil samples (the Florida Soil Characterization Database) while incorporating statistical methods to inform predictor selection for the model inputs. Of the machine learning (ML) models tested, random forest regression (RF) and gradient-boosted regression (GB) gave the best performances, with R2 = 0.71 and RMSE = 0.47 cm h-1 on the test data for both. Using the permutation feature importance technique, the GB and RF regression models showed similar results, where clay content described the most variation in the data, followed by bulk density. The implication of this study is that, when predicting Ks using the Florida Soil Characterization Database, priority should be given to obtaining quality data on clay content and bulk density as they are the most influential predictors for estimating Ks. Keywords: Deep learning, Gradient boosted regression, Pedotransfer functions, Random forest regression, Soil database, Soil properties.
在这项研究中，利用一个大型土壤数据库开发了六个机器学习(ML)模型，利用易于测量的土壤特征来预测这些土壤的饱和水力传导性。基于树的回归模型优于所有其他测试的ML模型。神经网络不适合预测饱和水导率。粘土含量，其次是体积密度，解释了所检查的预测数据中最大的变化。摘要其中一个最重要的土壤水力性质的模拟水在渗透带是饱和水力传导性。然而，在现场测量它是具有挑战性的。土壤传递函数(PTFs)是一种数学模型，可以根据容易测量的土壤特性预测饱和水力传导率(Ks)。虽然用于预测k的ptf的发展并不新鲜，但用于预测k的工具和方法仍在不断发展。模型的性能取决于选择用最少的输入变量解释最大数量的k方差的土壤特征。此外，大多数“黑箱”机器学习模型缺乏可解释性，这使得提取实用知识变得困难，因为机器学习过程模糊了PTF模型中输入和输出之间的关系。本研究的目的是开发一套新的ptf，用于使用机器学习算法和超过8000个土壤样本的大型数据库(佛罗里达土壤特征数据库)来预测k，同时结合统计方法来为模型输入的预测器选择提供信息。在测试的机器学习(ML)模型中，随机森林回归(RF)和梯度增强回归(GB)的性能最好，两者的测试数据的R2 = 0.71, RMSE = 0.47 cm h-1。使用排列特征重要性技术，GB和RF回归模型显示了相似的结果，其中粘土含量描述了数据中最大的变化，其次是体积密度。本研究的含义是，当使用佛罗里达土壤特征数据库预测k时，应优先考虑获得粘土含量和容重的高质量数据，因为它们是估计k的最具影响力的预测因子。关键词:深度学习，梯度增强回归，土壤传递函数，随机森林回归，土壤数据库，土壤性质
{"title":"Predictor Selection and Machine Learning Regression Methods to Predict Saturated Hydraulic Conductivity From a Large Public Soil Database","authors":"Toby A. Adjuik, S. Nokes, M. Montross, M. Sama, O. Wendroth","doi":"10.13031/ja.15068","DOIUrl":"https://doi.org/10.13031/ja.15068","url":null,"abstract":"Highlights In this study, six machine learning (ML) models were developed using a large database of soils to predict saturated hydraulic conductivity of these soils using easily measured soil characteristics. Tree-based regression models outperformed all other ML models tested. Neural networks were not suitable for predicting saturated hydraulic conductivity. Clay content, followed by bulk density, explained the highest amount of variation in the data of the predictors examined. Abstract. One of the most important soil hydraulic properties for modeling water transport in the vadose zone is saturated hydraulic conductivity. However, it is challenging to measure it in the field. Pedotransfer Functions (PTFs) are mathematical models that can predict saturated hydraulic conductivity (Ks) from easily measured soil characteristics. Though the development of PTFs for predicting Ks is not new, the tools and methods used to predict Ks are continuously evolving. Model performance depends on choosing soil features that explain the largest amount of Ks variance with the fewest input variables. In addition, the lack of interpretability in most “black box” machine learning models makes it difficult to extract practical knowledge as the machine learning process obfuscates the relationship between inputs and outputs in the PTF models. The objective of this study was to develop a set of new PTFs for predicting Ks using machine learning algorithms and a large database of over 8000 soil samples (the Florida Soil Characterization Database) while incorporating statistical methods to inform predictor selection for the model inputs. Of the machine learning (ML) models tested, random forest regression (RF) and gradient-boosted regression (GB) gave the best performances, with R2 = 0.71 and RMSE = 0.47 cm h-1 on the test data for both. Using the permutation feature importance technique, the GB and RF regression models showed similar results, where clay content described the most variation in the data, followed by bulk density. The implication of this study is that, when predicting Ks using the Florida Soil Characterization Database, priority should be given to obtaining quality data on clay content and bulk density as they are the most influential predictors for estimating Ks. Keywords: Deep learning, Gradient boosted regression, Pedotransfer functions, Random forest regression, Soil database, Soil properties.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"49 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79780815","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Highlights The automatic detection and recognition of farming action in video are realized. The YOLOv7-tiny was enhanced by incorporating Coordinate Attention (CA). The performance indices mAP@.5 and mAP@.5:.95 improved by 0.1% and 6.6%, respectively. An intelligent method for detecting "inspection" and "applying pesticides" is provided. Abstract. In aquaculture, regular "inspection" and "applying pesticides" are essential to improving production efficiency and fish disease treatment, but the current aquaculture system does not effectively support these strategies. Therefore, this paper proposes a farming action recognition network (FARnet), which can accurately locate the farmers in the video and detect the actions of “applying pesticides” and “inspection.” The dataset was captured and produced by multi-angle cameras, which were consulted with relevant experts. In this network, Coordinate Attention (CA) was used to improve the Efficient Layer Aggregation Networks-tiny (ELAN-tiny) and Spatial Pyramid Pooling (SPP) structures in the YOLOv7-tiny network. The precise implementation methods are as follows: (1) The convolution in ELAN-tiny was replaced with the CA module, and a shortcut was added. (2) A CA module was added to the final layer of the Spatial Pyramid Pooling (SPP) module. (3) The improved Efficient Layer Aggregation Networks-Coordinate Attention (ELAN-CA) and Spatial Pyramid Pooling-Coordinate Attention (SPP-CA) were used to extract action features and perform feature correction by ADD (Feature fusion by feature map summation) in the backbone. The results demonstrated that the FARnet achieved significantly better detection results than the YOLOv7-tiny network, where mAP@.5 improved by 0.1% from 99.4% to 99.5%, and the mAP@.5:.95 improved by 6.6% from 78.2% to 84.8%. Therefore, the FARnet can effectively detect and identify the “inspection” and “applying pesticides” actions of farmers and provide useful input information for the intelligent management system. Keywords: Action detection, Applying pesticides, Coordinate attention, FARnet, Inspection.
{"title":"FARnet: Farming Action Recognition From Videos Based on Coordinate Attention and YOLOv7-tiny Network in Aquaculture","authors":"Xinting Yang, Liang Pan, Dinghong Wang, Yuhao Zeng, Wentao Zhu, Dongxiang Jiao, Zhenlong Sun, Chuanheng Sun, Chao Zhou","doi":"10.13031/ja.15362","DOIUrl":"https://doi.org/10.13031/ja.15362","url":null,"abstract":"Highlights The automatic detection and recognition of farming action in video are realized. The YOLOv7-tiny was enhanced by incorporating Coordinate Attention (CA). The performance indices mAP@.5 and mAP@.5:.95 improved by 0.1% and 6.6%, respectively. An intelligent method for detecting \"inspection\" and \"applying pesticides\" is provided. Abstract. In aquaculture, regular \"inspection\" and \"applying pesticides\" are essential to improving production efficiency and fish disease treatment, but the current aquaculture system does not effectively support these strategies. Therefore, this paper proposes a farming action recognition network (FARnet), which can accurately locate the farmers in the video and detect the actions of “applying pesticides” and “inspection.” The dataset was captured and produced by multi-angle cameras, which were consulted with relevant experts. In this network, Coordinate Attention (CA) was used to improve the Efficient Layer Aggregation Networks-tiny (ELAN-tiny) and Spatial Pyramid Pooling (SPP) structures in the YOLOv7-tiny network. The precise implementation methods are as follows: (1) The convolution in ELAN-tiny was replaced with the CA module, and a shortcut was added. (2) A CA module was added to the final layer of the Spatial Pyramid Pooling (SPP) module. (3) The improved Efficient Layer Aggregation Networks-Coordinate Attention (ELAN-CA) and Spatial Pyramid Pooling-Coordinate Attention (SPP-CA) were used to extract action features and perform feature correction by ADD (Feature fusion by feature map summation) in the backbone. The results demonstrated that the FARnet achieved significantly better detection results than the YOLOv7-tiny network, where mAP@.5 improved by 0.1% from 99.4% to 99.5%, and the mAP@.5:.95 improved by 6.6% from 78.2% to 84.8%. Therefore, the FARnet can effectively detect and identify the “inspection” and “applying pesticides” actions of farmers and provide useful input information for the intelligent management system. Keywords: Action detection, Applying pesticides, Coordinate attention, FARnet, Inspection.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"10 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79931100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Highlights Terminal velocities were measured for wheat stem nodes and internodes for similar particle dimensions to investigate the feasibility of aerodynamic separation. Mean measures of terminal velocities for wheat stem nodes and internodes were 4.91 and 3.35 m s-1, respectively, that coincided with values of 4.92 and 3.37 m s-1 calculated for spherical particles (Mohsenin, 1970). Wheat stem particle mass ranged from 0.015 (internode) to 0.041 g (node) that significantly correlated with terminal velocity ranging from 3.13 to 5.14 m s-1, respectively. Wheat stem particle density ranged from 112 to 297 kg m-3 that significantly correlated with terminal velocity ranging from 3.12 to 5.11 m s-1, respectively. Abstract. Efficient separation of physiological plant components potentially improved the targeting of components to best uses. The terminal velocity property used an opposing air velocity to equilibrate particle weight with the sum of the drag and buoyancy forces. This study used particles of similar dimensions to ascertain the effect of particle mass and density on experimental measures of terminal velocity in a wind tunnel and as calculated by reliable equations. Similar particle diameters, lengths, and volumes of wheat stems ranged from 0.362 to 0.376 cm, 1.25 to 1.28 cm, and 0.131 to 0.141 cm3, respectively. Moisture content was 12% wet basis. Wheat stem internodes had individual particle mass and density ranging from 0.015 to 0.019 g and 113 to 144 kg m-3, respectively, and mean Terminal Velocity Wind Tunnel (TVWT) terminal velocities for wheat stem internodes that ranged from 3.13 to 3.58 m s-1. Nodes had individual particle mass and density ranging from 0.031 to 0.041 g and 236 to 297 kg m-3, respectively, and mean TVWT terminal velocities for wheat stem nodes that ranged from 4.62 to 5.14 m s-1. Thus, no overlap in values was observed for particle mass, particle density, and terminal velocity between wheat stem internode and wheat stem node. This observation supports the potential of using terminal velocity to separate node from internode for similar-sized wheat stems at a given moisture content. Keywords: Aerodynamic separation, Anatomical component, Biomass property, Physical experiment, Sorting, Terminal velocity, Vertical wind tunnel, Wheat stem particles.
为了探讨空气动力分离的可行性，对小麦茎节和茎节间相似颗粒尺寸的末端速度进行了测量。小麦茎节和节间的终端速度平均值分别为4.91和3.35 m s-1，这与球形颗粒的计算值4.92和3.37 m s-1相吻合(Mohsenin, 1970)。小麦茎粒质量在0.015 ~ 0.041 g(节)之间，与末速在3.13 ~ 5.14 m s-1之间显著相关。小麦茎秆颗粒密度在112 ~ 297 kg m-3之间，与终速在3.12 ~ 5.11 m s-1之间显著相关。摘要植物生理成分的有效分离有可能提高成分的靶向性。终端速度特性使用相反的空气速度来平衡颗粒重量与阻力和浮力的总和。本研究使用相似尺寸的粒子来确定粒子质量和密度对风洞中终端速度的实验测量的影响，并通过可靠的方程计算。小麦茎的相似粒径、长度和体积分别为0.362 ~ 0.376 cm、1.25 ~ 1.28 cm和0.131 ~ 0.141 cm3。水分含量为12%湿基。小麦茎秆节间的个体颗粒质量和密度分别为0.015 ~ 0.019 g和113 ~ 144 kg m-3，茎秆节间的平均终端风速(TVWT)为3.13 ~ 3.58 m s-1。小麦茎秆节点的粒子质量和密度分别为0.031 ~ 0.041 g和236 ~ 297 kg m-3，平均TVWT终端速度为4.62 ~ 5.14 m s-1。因此，小麦茎秆节间和茎秆节间的颗粒质量、颗粒密度和终端速度值没有重叠。这一观察结果支持了在一定含水量条件下，利用末端速度将类似大小的小麦茎节与节间分离的可能性。关键词:气动分离，解剖组分，生物量特性，物理实验，分选，终端速度，垂直风洞，小麦茎秆颗粒
{"title":"Terminal Velocity of Wheat Stem Nodes versus Internodes for Similar Particle Dimensions","authors":"A. Womac, S. E. Klasek, D. Yoder, Doug G. Hayes","doi":"10.13031/ja.15580","DOIUrl":"https://doi.org/10.13031/ja.15580","url":null,"abstract":"Highlights Terminal velocities were measured for wheat stem nodes and internodes for similar particle dimensions to investigate the feasibility of aerodynamic separation. Mean measures of terminal velocities for wheat stem nodes and internodes were 4.91 and 3.35 m s-1, respectively, that coincided with values of 4.92 and 3.37 m s-1 calculated for spherical particles (Mohsenin, 1970). Wheat stem particle mass ranged from 0.015 (internode) to 0.041 g (node) that significantly correlated with terminal velocity ranging from 3.13 to 5.14 m s-1, respectively. Wheat stem particle density ranged from 112 to 297 kg m-3 that significantly correlated with terminal velocity ranging from 3.12 to 5.11 m s-1, respectively. Abstract. Efficient separation of physiological plant components potentially improved the targeting of components to best uses. The terminal velocity property used an opposing air velocity to equilibrate particle weight with the sum of the drag and buoyancy forces. This study used particles of similar dimensions to ascertain the effect of particle mass and density on experimental measures of terminal velocity in a wind tunnel and as calculated by reliable equations. Similar particle diameters, lengths, and volumes of wheat stems ranged from 0.362 to 0.376 cm, 1.25 to 1.28 cm, and 0.131 to 0.141 cm3, respectively. Moisture content was 12% wet basis. Wheat stem internodes had individual particle mass and density ranging from 0.015 to 0.019 g and 113 to 144 kg m-3, respectively, and mean Terminal Velocity Wind Tunnel (TVWT) terminal velocities for wheat stem internodes that ranged from 3.13 to 3.58 m s-1. Nodes had individual particle mass and density ranging from 0.031 to 0.041 g and 236 to 297 kg m-3, respectively, and mean TVWT terminal velocities for wheat stem nodes that ranged from 4.62 to 5.14 m s-1. Thus, no overlap in values was observed for particle mass, particle density, and terminal velocity between wheat stem internode and wheat stem node. This observation supports the potential of using terminal velocity to separate node from internode for similar-sized wheat stems at a given moisture content. Keywords: Aerodynamic separation, Anatomical component, Biomass property, Physical experiment, Sorting, Terminal velocity, Vertical wind tunnel, Wheat stem particles.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"3 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"82458995","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Te Xi, Lunqing Sun, Yongwei Wang, Dong-Lin Li, Fake Shanno, Fuqiang Yao, Jun Wang
Highlights The effect of airflow velocity on pollen distribution was investigated under a large-scale planting mode. The response surface model between pollen distribution and airflow velocity was constructed. Multi-objective optimization of airflow velocity combinations was carried out using a genetic algorithm. The optimal airflow velocity ranges of the male parents are from 22.4 to 24 m/s, 23.1 to 27 m/s, and 23.5 to 24.1 m/s. Abstract. Pollination is the key link in hybrid rice seed production. The pneumatic pollination method can significantly improve pollination efficiency under large-scale planting mode. To investigate the effect of airflow velocity on pollen distribution in hybrid rice pollination, the velocities of airflow acting on different male parent rows were taken as the experimental factors. The pollen amount in per view and the variation rate of pollen distribution in female parent rows were used as experimental indices. Field experiments were carried out using a self-made pneumatic pollination experimental platform. The results showed that when the airflow acted on the male parents in the first and second rows of the adjacent female parent, the pollen dissemination distance was short when the airflow velocity was low, and the pollen was mainly deposited in the area near the male parents. With the increase in airflow velocity, the peak pollen amount in per view in the female parent rows gradually moved away from the male parent rows. But they are all in the female parent rows of the effective area. The total amount of pollen also increased. Due to the blocking effect of the outer male parent row, the pollen dissemination was restricted when the airflow alone acted on the third male parent row. The effect of airflow velocity on pollen distribution was not obvious. The experimental results of different airflow velocities acting on the parent row alone are used as the basis. The objective functions of pollen amount, distribution variation rate, and airflow velocities of each male parent row were established by response surface methodology. The multi-objective optimization of airflow velocity combinations was carried out by a genetic algorithm. The pollen distribution under different air velocity combinations was obtained. When the optimal airflow velocity ranges of the male parents in rows 1, 2, and 3 are 22.4 to 24 m/s, 23.1 to 27 m/s, and 23.5 to 24.1 m/s, respectively, pollination is uniform and sufficient. The research results can provide a basis for the development of pneumatic pollinators and the optimization of working parameters under large-scale planting mode. Keywords: Multi-objective parameter optimization, Pneumatic pollination machinery, Response surface modeling, Rice pollination.
{"title":"Optimizing the Airflow Velocity Combinations Acting on Male Parent Rows for Hybrid Rice Pollination","authors":"Te Xi, Lunqing Sun, Yongwei Wang, Dong-Lin Li, Fake Shanno, Fuqiang Yao, Jun Wang","doi":"10.13031/ja.15233","DOIUrl":"https://doi.org/10.13031/ja.15233","url":null,"abstract":"Highlights The effect of airflow velocity on pollen distribution was investigated under a large-scale planting mode. The response surface model between pollen distribution and airflow velocity was constructed. Multi-objective optimization of airflow velocity combinations was carried out using a genetic algorithm. The optimal airflow velocity ranges of the male parents are from 22.4 to 24 m/s, 23.1 to 27 m/s, and 23.5 to 24.1 m/s. Abstract. Pollination is the key link in hybrid rice seed production. The pneumatic pollination method can significantly improve pollination efficiency under large-scale planting mode. To investigate the effect of airflow velocity on pollen distribution in hybrid rice pollination, the velocities of airflow acting on different male parent rows were taken as the experimental factors. The pollen amount in per view and the variation rate of pollen distribution in female parent rows were used as experimental indices. Field experiments were carried out using a self-made pneumatic pollination experimental platform. The results showed that when the airflow acted on the male parents in the first and second rows of the adjacent female parent, the pollen dissemination distance was short when the airflow velocity was low, and the pollen was mainly deposited in the area near the male parents. With the increase in airflow velocity, the peak pollen amount in per view in the female parent rows gradually moved away from the male parent rows. But they are all in the female parent rows of the effective area. The total amount of pollen also increased. Due to the blocking effect of the outer male parent row, the pollen dissemination was restricted when the airflow alone acted on the third male parent row. The effect of airflow velocity on pollen distribution was not obvious. The experimental results of different airflow velocities acting on the parent row alone are used as the basis. The objective functions of pollen amount, distribution variation rate, and airflow velocities of each male parent row were established by response surface methodology. The multi-objective optimization of airflow velocity combinations was carried out by a genetic algorithm. The pollen distribution under different air velocity combinations was obtained. When the optimal airflow velocity ranges of the male parents in rows 1, 2, and 3 are 22.4 to 24 m/s, 23.1 to 27 m/s, and 23.5 to 24.1 m/s, respectively, pollination is uniform and sufficient. The research results can provide a basis for the development of pneumatic pollinators and the optimization of working parameters under large-scale planting mode. Keywords: Multi-objective parameter optimization, Pneumatic pollination machinery, Response surface modeling, Rice pollination.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"47 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81353221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jizhong Wang, Yangchun Liu, Bo Zhao, Fengzhu Wang, Weipeng Zhang, Yang Li
Highlights Prevent metal foreign bodies from scratching the intestines of animals and damaging the harvest cutter. Highly integrated design of acquisition circuit. Application of electromagnetic simulation to verify the feasibility of the principle of eddy current effect. Establishment of Support Vector Machine Multi-Classification Algorithm Model. Abstract. Aiming at the problem that the metal foreign bodies mingled in the silage cause damage to the gastrointestinal tract of animals and livestock, as well as irreversible damage to the rotary cutter of the harvester, a metal foreign body detection and sensing device for the harvester feeding drum composed of multiple single coils and signal acquisition units was designed to realize real-time detection and alarm of metal foreign bodies during harvesting. The sensor adopted a monolithic design with high integration of the signal acquisition circuit, which has a strong anti-interference ability. First, the electromagnetic simulation model was established. According to the simulation analysis of the eddy current effect, when the metal foreign object enters the alternating magnetic field, the energy will be lost, and the equivalent impedance of the coil will change accordingly. Then, the existence of the metal foreign body can be determined by detecting the equivalent impedance Rp of the coil. Next, we adopted a support vector machine multi-classification algorithm to train the detection device. In this process, different sizes of metal (copper, aluminum, and iron) were used, which can effectively improve the sensitivity and accuracy of metal foreign body detection. Finally, the sensor was installed on the test stand for multi-scene simulation experiments. The results show that the metal detection sensor can quickly identify the existence of metal by detecting the equivalent impedance Rp based on the eddy current effect; at the same time, the size of this sensor for metal foreign body detection is limited to 0.6 mm in diameter, 12 mm in length, and 100 mm in maximum detecting distance. Keywords: Eddy current effect, Equivalent impedance, Harvester, Metal foreign body, Support vector machine.
{"title":"Design and Verification of Metal Foreign Body Detection Device for Harvester Based on Eddy Current Effect","authors":"Jizhong Wang, Yangchun Liu, Bo Zhao, Fengzhu Wang, Weipeng Zhang, Yang Li","doi":"10.13031/ja.15185","DOIUrl":"https://doi.org/10.13031/ja.15185","url":null,"abstract":"Highlights Prevent metal foreign bodies from scratching the intestines of animals and damaging the harvest cutter. Highly integrated design of acquisition circuit. Application of electromagnetic simulation to verify the feasibility of the principle of eddy current effect. Establishment of Support Vector Machine Multi-Classification Algorithm Model. Abstract. Aiming at the problem that the metal foreign bodies mingled in the silage cause damage to the gastrointestinal tract of animals and livestock, as well as irreversible damage to the rotary cutter of the harvester, a metal foreign body detection and sensing device for the harvester feeding drum composed of multiple single coils and signal acquisition units was designed to realize real-time detection and alarm of metal foreign bodies during harvesting. The sensor adopted a monolithic design with high integration of the signal acquisition circuit, which has a strong anti-interference ability. First, the electromagnetic simulation model was established. According to the simulation analysis of the eddy current effect, when the metal foreign object enters the alternating magnetic field, the energy will be lost, and the equivalent impedance of the coil will change accordingly. Then, the existence of the metal foreign body can be determined by detecting the equivalent impedance Rp of the coil. Next, we adopted a support vector machine multi-classification algorithm to train the detection device. In this process, different sizes of metal (copper, aluminum, and iron) were used, which can effectively improve the sensitivity and accuracy of metal foreign body detection. Finally, the sensor was installed on the test stand for multi-scene simulation experiments. The results show that the metal detection sensor can quickly identify the existence of metal by detecting the equivalent impedance Rp based on the eddy current effect; at the same time, the size of this sensor for metal foreign body detection is limited to 0.6 mm in diameter, 12 mm in length, and 100 mm in maximum detecting distance. Keywords: Eddy current effect, Equivalent impedance, Harvester, Metal foreign body, Support vector machine.","PeriodicalId":29714,"journal":{"name":"Journal of the ASABE","volume":"32 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87146569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","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：