Wheat stripe rust poses a marked threat to global wheat production.Accurate and effective disease severity assessments are crucial for disease resistance breeding and timely management of field diseases.In this study,...Wheat stripe rust poses a marked threat to global wheat production.Accurate and effective disease severity assessments are crucial for disease resistance breeding and timely management of field diseases.In this study,we propose a practical solution using mobile-based deep learning and model-assisted labeling.StripeRust-Pocket,a user-friendly mobile application developed based on deep learning models,accurately quantifies disease severity in wheat stripe rust leaf images,even under complex backgrounds.Additionally,StripeRust-Pocket facilitates image acquisition,result storage,organization,and sharing.The underlying model employed by StripeRust-Pocket,called StripeRustNet,is a balanced lightweight 2-stage model.The first stage utilizes MobileNetV2-DeepLabV3+for leaf segmentation,followed by ResNet50-DeepLabV3+in the second stage for lesion segmentation.Disease severity is estimated by calculating the ratio of the lesion pixel area to the leaf pixel area.StripeRustNet achieves 98.65%mean intersection over union(MIoU)for leaf segmentation and 86.08%MIoU for lesion segmentation.Validation using an additional 100 field images demonstrated a mean correlation of over 0.964 with 3 expert visual scores.To address the challenges in manual labeling,we introduce a 2-stage labeling pipeline that combines model-assisted labeling,manual correction,and spatial complementarity.We apply this pipeline to our self-collected dataset,reducing the annotation time from 20 min to 3 min per image.Our method provides an efficient and practical solution for wheat stripe rust severity assessments,empowering wheat breeders and pathologists to implement timely disease management.It also demonstrates how to address the"last mile"challenge of applying computer vision technology to plant phenomics.展开更多
Grape crops are a great source of income for farmers.The yield and quality of grapes can be improved by preventing and treating diseases.The farmer’s yield will be dramatically impacted if diseases are found on grape...Grape crops are a great source of income for farmers.The yield and quality of grapes can be improved by preventing and treating diseases.The farmer’s yield will be dramatically impacted if diseases are found on grape leaves.Automatic detection can reduce the chances of leaf diseases affecting other healthy plants.Several studies have been conducted to detect grape leaf diseases,but most fail to engage with end users and integrate the model with real-time mobile applications.This study developed a mobile-based grape leaf disease detection(GLDD)application to identify infected leaves,Grape Guard,based on a TensorFlow Lite(TFLite)model generated from the You Only Look Once(YOLO)v8 model.A public grape leaf disease dataset containing four classes was used to train the model.The results of this study were relied on the YOLO architecture,specifically YOLOv5 and YOLOv8.After extensive experiments with different image sizes,YOLOv8 performed better than YOLOv5.YOLOv8 achieved 99.9%precision,100%recall,99.5%mean average precision(mAP),and 88%mAP50-95 for all classes to detect grape leaf diseases.The Grape Guard android mobile application can accurately detect the grape leaf disease by capturing images from grape vines.展开更多
Among the different types of traffic sign damage, vandalism is exclusively caused by humans. Traffic sign vandalism is a serious concern, since it can lead to an increase in unsafe driving behaviors. In addition, it r...Among the different types of traffic sign damage, vandalism is exclusively caused by humans. Traffic sign vandalism is a serious concern, since it can lead to an increase in unsafe driving behaviors. In addition, it results in increased costs to transportation agencies to replace, repair, or maintain the vandalized signs. This paper examines the association between the local population demographics and traffic sign vandalism rates in the State of Utah. To accomplish this goal, sign data of over 97,000 traffic signs across Utah were digitally collected by an equipped vehicle. Sign damage data were obtained from the inspection of daytime digital images taken of each individual sign. Demographic data of Utah＇s counties, including population density, ethnicity, age, income, education, and gender, were obtained from the U.S. Census. The association between demographic groups and vandalism rates was tested using chi-square and trend tests. The results reveal that the most statistically significant variables comprise median household income, completion of at least an associate degree, and population density. According to the fitted linear regression model, a relationship exists between sign vandalism rate and local population demographic. The findings of this investigation can assist transportation agencies in identifying areas with a higher likelihood of sign vandalism, based on demographic characteristics. Such information can then be used to encourage scheduled sign inspections and to implement various countermeasures to prevent sign vandalism.展开更多
基金partially supported by the National Natural Science Foundation of China(Grant Nos.32200331 and 32090061)the Major Science and Technology Research Project of Hubei Province(Grant No.2021 AFB002)+1 种基金the Natural Science Foundation of Chongqing(Grant No.cstc2021jcyj-msxmX1050)the Open Project of Wuhan University of Technology Chongqing Research Institute(Grant No.ZL2021-3).
文摘Wheat stripe rust poses a marked threat to global wheat production.Accurate and effective disease severity assessments are crucial for disease resistance breeding and timely management of field diseases.In this study,we propose a practical solution using mobile-based deep learning and model-assisted labeling.StripeRust-Pocket,a user-friendly mobile application developed based on deep learning models,accurately quantifies disease severity in wheat stripe rust leaf images,even under complex backgrounds.Additionally,StripeRust-Pocket facilitates image acquisition,result storage,organization,and sharing.The underlying model employed by StripeRust-Pocket,called StripeRustNet,is a balanced lightweight 2-stage model.The first stage utilizes MobileNetV2-DeepLabV3+for leaf segmentation,followed by ResNet50-DeepLabV3+in the second stage for lesion segmentation.Disease severity is estimated by calculating the ratio of the lesion pixel area to the leaf pixel area.StripeRustNet achieves 98.65%mean intersection over union(MIoU)for leaf segmentation and 86.08%MIoU for lesion segmentation.Validation using an additional 100 field images demonstrated a mean correlation of over 0.964 with 3 expert visual scores.To address the challenges in manual labeling,we introduce a 2-stage labeling pipeline that combines model-assisted labeling,manual correction,and spatial complementarity.We apply this pipeline to our self-collected dataset,reducing the annotation time from 20 min to 3 min per image.Our method provides an efficient and practical solution for wheat stripe rust severity assessments,empowering wheat breeders and pathologists to implement timely disease management.It also demonstrates how to address the"last mile"challenge of applying computer vision technology to plant phenomics.
文摘Grape crops are a great source of income for farmers.The yield and quality of grapes can be improved by preventing and treating diseases.The farmer’s yield will be dramatically impacted if diseases are found on grape leaves.Automatic detection can reduce the chances of leaf diseases affecting other healthy plants.Several studies have been conducted to detect grape leaf diseases,but most fail to engage with end users and integrate the model with real-time mobile applications.This study developed a mobile-based grape leaf disease detection(GLDD)application to identify infected leaves,Grape Guard,based on a TensorFlow Lite(TFLite)model generated from the You Only Look Once(YOLO)v8 model.A public grape leaf disease dataset containing four classes was used to train the model.The results of this study were relied on the YOLO architecture,specifically YOLOv5 and YOLOv8.After extensive experiments with different image sizes,YOLOv8 performed better than YOLOv5.YOLOv8 achieved 99.9%precision,100%recall,99.5%mean average precision(mAP),and 88%mAP50-95 for all classes to detect grape leaf diseases.The Grape Guard android mobile application can accurately detect the grape leaf disease by capturing images from grape vines.
文摘Among the different types of traffic sign damage, vandalism is exclusively caused by humans. Traffic sign vandalism is a serious concern, since it can lead to an increase in unsafe driving behaviors. In addition, it results in increased costs to transportation agencies to replace, repair, or maintain the vandalized signs. This paper examines the association between the local population demographics and traffic sign vandalism rates in the State of Utah. To accomplish this goal, sign data of over 97,000 traffic signs across Utah were digitally collected by an equipped vehicle. Sign damage data were obtained from the inspection of daytime digital images taken of each individual sign. Demographic data of Utah＇s counties, including population density, ethnicity, age, income, education, and gender, were obtained from the U.S. Census. The association between demographic groups and vandalism rates was tested using chi-square and trend tests. The results reveal that the most statistically significant variables comprise median household income, completion of at least an associate degree, and population density. According to the fitted linear regression model, a relationship exists between sign vandalism rate and local population demographic. The findings of this investigation can assist transportation agencies in identifying areas with a higher likelihood of sign vandalism, based on demographic characteristics. Such information can then be used to encourage scheduled sign inspections and to implement various countermeasures to prevent sign vandalism.
