Precision agriculture(PA)is an agricultural management strategy based on observation,measurement and response to the variability of inter/intra-champ cultures.It includes advances in terms of data collection,analysis ...Precision agriculture(PA)is an agricultural management strategy based on observation,measurement and response to the variability of inter/intra-champ cultures.It includes advances in terms of data collection,analysis and management,as well as technological developments in terms of data storage and recovery,precise positioning,yield monitoring and remote sensing.The latter offers an unprecedented spatial,spectral and temporal resolution,but can also provide detailed information on the height of the vegetation and various observations.Today,the success of new agricultural technologies means that many agricultural tasks have become automated and that scientists have conducted more studies and research based on smart algorithms that automatically learn the decision rules from data.The use of Deep Learning(DL)and in particular the development and application of some of its algorithms called Convolutional Neural Networks(CNNs)are considered to be a particular success.In this work,we have applied and tested the performance of a network of convolutional neural network to automatically detect and map olive trees from Phantom4 drone imagery.The workflow involved the acquisition of images and the generation of ortho-mosaic with Pix4D software,as well as the use of a geographic information system.The results obtained with a training dataset of 4500 images of 24∗24 pixels are very satisfying:95%Precision,a 99%Recall and an F-score of 97%.展开更多
The harvesting process of the olive tree is mainly performed by manual means,because traditional olive orchards(the main planting typology)are formed of irregular,large-canopy trees that are very difficult to harvest ...The harvesting process of the olive tree is mainly performed by manual means,because traditional olive orchards(the main planting typology)are formed of irregular,large-canopy trees that are very difficult to harvest mechanically.For that reason,the cost of harvesting is very high,and it threatens the future of these plantations whose conversion to other more modern layouts is not always possible due to several limitations.The introduction of a harvester may represent the technological change that is the key factor for improved competitiveness.The main purpose of this work was to develop a harvester based on canopy shaker technology for work on irregular,large trees in a circular path.The design of the harvester was based on a determination of tree geometry,together with tree training.Field tests were used to determine machine-tree interaction,and to evaluate the removal,catch frame and driven systems.The proposed innovation allowed the fully mechanical harvest of previously planted trees with a removal efficiency of over 84%,achieving an effective field capacity of 0.21 hm^(2)/h.Although the results so far have been promising,further improvements are advisable in machine and tree adaptation.展开更多
文摘Precision agriculture(PA)is an agricultural management strategy based on observation,measurement and response to the variability of inter/intra-champ cultures.It includes advances in terms of data collection,analysis and management,as well as technological developments in terms of data storage and recovery,precise positioning,yield monitoring and remote sensing.The latter offers an unprecedented spatial,spectral and temporal resolution,but can also provide detailed information on the height of the vegetation and various observations.Today,the success of new agricultural technologies means that many agricultural tasks have become automated and that scientists have conducted more studies and research based on smart algorithms that automatically learn the decision rules from data.The use of Deep Learning(DL)and in particular the development and application of some of its algorithms called Convolutional Neural Networks(CNNs)are considered to be a particular success.In this work,we have applied and tested the performance of a network of convolutional neural network to automatically detect and map olive trees from Phantom4 drone imagery.The workflow involved the acquisition of images and the generation of ortho-mosaic with Pix4D software,as well as the use of a geographic information system.The results obtained with a training dataset of 4500 images of 24∗24 pixels are very satisfying:95%Precision,a 99%Recall and an F-score of 97%.
基金The authors gratefully acknowledge financial support from the Spanish Ministry of Economy and Competitiveness(PCP Mecaolivar),and the Spanish Olive Oil Interprofessional Organisation.
文摘The harvesting process of the olive tree is mainly performed by manual means,because traditional olive orchards(the main planting typology)are formed of irregular,large-canopy trees that are very difficult to harvest mechanically.For that reason,the cost of harvesting is very high,and it threatens the future of these plantations whose conversion to other more modern layouts is not always possible due to several limitations.The introduction of a harvester may represent the technological change that is the key factor for improved competitiveness.The main purpose of this work was to develop a harvester based on canopy shaker technology for work on irregular,large trees in a circular path.The design of the harvester was based on a determination of tree geometry,together with tree training.Field tests were used to determine machine-tree interaction,and to evaluate the removal,catch frame and driven systems.The proposed innovation allowed the fully mechanical harvest of previously planted trees with a removal efficiency of over 84%,achieving an effective field capacity of 0.21 hm^(2)/h.Although the results so far have been promising,further improvements are advisable in machine and tree adaptation.
