Skip to main content

DESARROLLO Y VALIDACION DE NUEVAS TECNOLOGIAS DE TELEDETECCION Y APRENDIZAJE AUTOMATICO APLICADAS AL CONTROL INTELIGENTE DE MALAS HIERBAS

PID2020-113229RB-C41

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Call Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Call year 2020
State Plan 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre AGENCIA ESTATAL CONSEJO SUPERIOR DE INVESTIGACIONES CIENTIFICAS (CSIC)

Resultados relacionados

Resultados totales (Incluyendo duplicados): 30
Encontrada(s) 1 página(s)

Repositori Obert UdL
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters - Part 2: Comparison for different crops and training systems

  • Torres-Sánchez, Jorge
  • Escolà i Agustí, Alexandre
  • de Castro, Ana I.
  • López-Granados, Francisca
  • Rosell Polo, Joan Ramon
  • Sebé Feixas, Francesc
  • Jiménez-Brenes, Francisco M.
  • Sanz Cortiella, Ricardo
  • Gregorio López, Eduard
  • Peña, José M.
The measurement of geometric canopy parameters in woody crops is an important task in Precision Agriculture because of their correlation with crop condition and productivity. In recent years, several technological approaches have been developed as an alternative to manual measurements, which are time- and labour-consuming. Two of the most commonly used 3D canopy characterization technologies are mobile terrestrial laser scanning (MTLS) based on light detection and ranging (LiDAR) sensors, and digital aerial photogrammetry (DAP) using imagery from uncrewed aerial vehicles (UAVs). Although both are state-of-the-art and have been fully tested and validated, a complete comparison between their geometric canopy parameter estimations in different woody crops and training systems has not been carried out. For this reason, a set of geometric parameters (canopy height, projected area, and volume) of a vineyard, an intensive peach orchard, and an intensive pear orchard were measured using UAV-DAP and MTLS-LiDAR. A comparison between both kinds of measurements was performed, accounting for the length of the sections in which the crop hedgerows were divided to extract the geometric parameters. Measurements from the UAV and the MTLS were highly correlated (R2 from 0.82 to 0.94) when considering the data from the three crops together, and the correlations were higher when analysing longer row sections. The canopy geometric parameters estimated using the MTLS-LiDAR always had higher values than those from the UAV-DAP. The results presented in this work provide useful data for a more informed selection of technological approaches for 3D crop characterization in Precision Fruticulture and high-throughput phenotyping., This research was funded by the projects DECIMAL (PID2020-113229RB-C41), PRODIGIA (PID2020-113229RB-C44), AGVANCE (AGL2013-48297-C2-2-R), and PAgPROTECT (PID2021-126648OB-I00), from the Spanish Ministry of Science and Innovation (MCIN/AEI/10.13039/501100011033). The authors also wish to thank Codorníu SA, celler Raimat and IRTA for having allowed the use of their vineyards, pear and peach orchards to conduct the trials described in this work.
Repositori Obert UdL
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters - Part 1: Methodology and comparison in vineyards

  • Escolà i Agustí, Alexandre
  • Peña, José M.
  • López-Granados, Francisca
  • Rosell Polo, Joan Ramon
  • de Castro, Ana I.
  • Gregorio López, Eduard
  • Jiménez-Brenes, Francisco M.
  • Sanz Cortiella, Ricardo
  • Sebé Feixas, Francesc
  • Llorens Calveras, Jordi
  • Torres-Sánchez, Jorge
Characterizing crop canopies is especially important in the management of woody crops. In this article, two systems were compared to characterise a 50 m long vineyard row section. One of the systems was a mobile terrestrial laser scanner based on a light detection and ranging (LiDAR) sensor (MTLS-LiDAR). The other was an uncrewed aerial vehicle (UAV) based system using digital aerial photogrammetry (UAV-DAP). The resulting 3D point clouds were assessed qualitatively and quantitatively. Canopy heights, widths and volumes were obtained in 0.1 m long sections along the studied row. All the parameters derived from the two systems presented statistically significant differences. The coefficients of determination between systems were 0.619 for canopy maximum heights above ground level (agl), 0.686 for 90th percentile (P90) heights agl, and 0.283 and 0.274 for maximum and P90 vegetated heights, respectively. Coefficients of determination between averaged maximum canopy width and P90 canopy width were 0.328 and 0.317, respectively. Coefficients of determination between cross-sectional areas determined from maximum widths, P90 widths and from the occupancy grid method were 0.423, 0.409 and 0.334, respectively. Total canopy volume for the entire row obtained from the three cross section estimation methods differed between 19 m3 and 25 m3. The reasons found were that the MTLS-LiDAR-derived point cloud captured the canopy top and side variability but could be affected by occlusions, mixed pixels and tall grass-like weeds present in the surveyed area. For its part, the UAV-DAP-derived point cloud tended to miss top and side shoots and somewhat smoothed canopy variability. As neither of the systems is optimal, a balance needs to be found according to the specific requirements of the survey. For this purpose, a list of pros and cons is presented to support the selection of one of the two systems for canopy monitoring. The MTLS-LiDAR system should be chosen when high detail is required but small areas are to be scanned. Alternatively, the UAV-DAP system should be chosen when large areas are to be monitored and when canopy detail is not so important. Further results are presented in Part 2 for a larger area and including pear and peach orchards with different training systems. Future research is to be conducted on how the compared systems affect variability detection and support variable-rate prescriptions., This research was funded by the projects AGVANCE (AGL2013-48297-C2-2-R), PAgPROTECT (PID2021-126648OB-I00), DECIMAL (PID2020-113229RB-C41) and PRODIGIA (PID2020-113229RB-C44) from the Spanish Ministry of Science and Innovation MICIN/AEI/10.13039/501100011033/FEDER, UE. The authors also want to thank Codorníu SA and Celler Raimat for allowing the use of one of their vineyards to conduct the trial described in this work.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Use of Under-Vine Living Mulches to Control Noxious Weeds in Irrigated Mediterranean Vineyards

  • Guerra, José G.
  • Cabello, Félix
  • Fernández-Quintanilla, César
  • Peña Barragán, José Manuel
  • Dorado, José
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/)., This article assesses the use of under-vine living mulches in Mediterranean vineyards characterized by limited water resources, one of the reasons why this agronomic practice is currently unusual in these environments. The aim of the study was to test whether the use of this alternative method in Mediterranean vineyards could suppress noxious weeds without hindering optimal vineyard development. For this purpose, four native species were selected as living mulches: Festuca ovina, Pilosella officinarum, Plantago coronopus, and Plantago lanceolata. The variables measured during three years in two different experimental farms were: (a) living mulch cover, as a possible predictor of weed suppression success; (b) weed density and weed biomass, with special attention to noxious weed species; and (c) pruning weights, measured in the last year to analyze the cumulative effect of the treatments on the grapevine vegetative growth. Our results revealed that living mulches with high cover rates (average over 70%) also showed weed suppression of up to 95%, significantly controlling the occurrence of noxious weeds such as Erigeron canadensis. No significant effect of the different treatments on vine vegetative growth was found, although further studies would be necessary. Based on these findings, it can be concluded that under-vine living mulches could be an efficient and environmentally friendly method for weed control in Mediterranean vineyards where irrigation is available., This research was funded by MICINN-AEI, grant number PID2020-113229RB-C41/AEI/10.13039/501100011033 and by FEDER/MICINN-AEI, grant number AGL2017-83325-C4-1-R. The lead author has been beneficiary of a predoctoral research grant FPI-INIA2016-00035., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Boosting precision crop protection towards agriculture 5.0 via machine learning and emerging technologies: A contextual review

  • Mesías-Ruiz, Gustavo A.
  • Pérez-Ortiz, María
  • Dorado, José
  • Castro, Ana Isabel de
  • Peña Barragán, José Manuel
© 2023 Mesías-Ruiz, Pérez-Ortiz, Dorado, de Castro and Peña. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms., Crop protection is a key activity for the sustainability and feasibility of agriculture in a current context of climate change, which is causing the destabilization of agricultural practices and an increase in the incidence of current or invasive pests, and a growing world population that requires guaranteeing the food supply chain and ensuring food security. In view of these events, this article provides a contextual review in six sections on the role of artificial intelligence (AI), machine learning (ML) and other emerging technologies to solve current and future challenges of crop protection. Over time, crop protection has progressed from a primitive agriculture 1.0 (Ag1.0) through various technological developments to reach a level of maturity closelyin line with Ag5.0 (section 1), which is characterized by successfully leveraging ML capacity and modern agricultural devices and machines that perceive, analyze and actuate following the main stages of precision crop protection (section 2). Section 3 presents a taxonomy of ML algorithms that support the development and implementation of precision crop protection, while section 4 analyses the scientific impact of ML on the basis of an extensive bibliometric study of >120 algorithms, outlining the most widely used ML and deep learning (DL) techniques currently applied in relevant case studies on the detection and control of crop diseases, weeds and plagues. Section 5 describes 39 emerging technologies in the fields of smart sensors and other advanced hardware devices, telecommunications, proximal and remote sensing, and AI-based robotics that will foreseeably lead the next generation of perception-based, decision-making and actuation systems for digitized, smart and real-time crop protection in a realistic Ag5.0. Finally, section 6 highlights the main conclusions and final remarks., This work was supported by the Spanish Research State Agency (AEI) through the Projects PDC2021-121537-C22/AEI/10.13039/501100011033 and PID2020-113229RB-C41. The lead author GM-R has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Early-Season Mapping of Johnsongrass (Sorghum halepense), Common Cocklebur (Xanthium strumarium) and Velvetleaf (Abutilon theophrasti) in Corn Fields Using Airborne Hyperspectral Imagery

  • Martin, M. Pilar
  • Ponce, Bernarda
  • Echavarría, Pilar
  • Dorado, José
  • Fernández-Quintanilla, César
Este artículo está sujeto a una licencia CC BY 4.0, Accurate information on the spatial distribution of weeds is the key to effective site-specific weed management and the efficient and sustainable use of weed control measures. This work focuses on the early detection of johnsongrass, common cocklebur and velvetleaf present in a corn field using high resolution airborne hyperspectral imagery acquired when corn plants were in a four to six leaf growth stage. Following the appropriate radiometric and geometric corrections, two supervised classification techniques, such as spectral angle mapper (SAM) and spectral mixture analysis (SMA) were applied. Two different procedures were compared for endmember selections: field spectral measurements and automatic methods to identify pure pixels in the image. Maps for both, overall weeds and for each of the three weed species, were obtained with the different classification methods and endmember sources. The best results were achieved by defining the endmembers through spectral information collected with a field spectroradiometer. Overall accuracies ranged between 60% and 80% using SAM for maps that do not differentiate the weed species while it decreased to 52% when the three weed species were individually classified. In this case, the SMA classification technique clearly improved the SAM results. The proposed methodology shows it to be a promising prospect to be applicable to low cost images acquired by the new generation of hyperspectral sensors onboard unmanned aerial vehicles (UAVs)., This research was supported by the Spanish Research State Agency (AEI) through the Project PID2020-113229RB-C41/AEI/10.13039/501100011033, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters – Part 2: Comparison for different crops and training systems

  • Torres-Sánchez, Jorge
  • Escolà Agustí, Alexandre
  • Castro, Ana Isabel de
  • López Granados, Francisca
  • Rosell-Polo, Joan R.
  • Sebé, Francesc
  • Jiménez-Brenes, Francisco Manuel
  • Sanz, Ricardo
  • Gregorio, Eduard
  • Peña Barragán, José Manuel
10 Pág., The measurement of geometric canopy parameters in woody crops is an important task in Precision Agriculture because of their correlation with crop condition and productivity. In recent years, several technological approaches have been developed as an alternative to manual measurements, which are time- and labour-consuming. Two of the most commonly used 3D canopy characterization technologies are mobile terrestrial laser scanning (MTLS) based on light detection and ranging (LiDAR) sensors, and digital aerial photogrammetry (DAP) using imagery from uncrewed aerial vehicles (UAVs). Although both are state-of-the-art and have been fully tested and validated, a complete comparison between their geometric canopy parameter estimations in different woody crops and training systems has not been carried out. For this reason, a set of geometric parameters (canopy height, projected area, and volume) of a vineyard, an intensive peach orchard, and an intensive pear orchard were measured using UAV-DAP and MTLS-LiDAR. A comparison between both kinds of measurements was performed, accounting for the length of the sections in which the crop hedgerows were divided to extract the geometric parameters. Measurements from the UAV and the MTLS were highly correlated (R2 from 0.82 to 0.94) when considering the data from the three crops together, and the correlations were higher when analysing longer row sections. The canopy geometric parameters estimated using the MTLS-LiDAR always had higher values than those from the UAV-DAP. The results presented in this work provide useful data for a more informed selection of technological approaches for 3D crop characterization in Precision Fruticulture and high-throughput phenotyping., This research was funded by the projects DECIMAL (PID2020-113229RB-C41), PRODIGIA (PID2020-113229RB-C44), AGVANCE (AGL2013-48297-C2-2-R), and PAgPROTECT (PID2021-126648OB-I00), from the Spanish Ministry of Science and Innovation (MCIN/AEI/10.13039/501100011033). The authors also wish to thank Codorníu SA, celler Raimat and IRTA for having allowed the use of their vineyards, pear and peach orchards to conduct the trials described in this work., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters – Part 1: Methodology and comparison in vineyards

  • Escolà Agustí, Alexandre
  • Peña Barragán, José Manuel
  • López Granados, Francisca
  • Rosell-Polo, Joan R.
  • Castro, Ana Isabel de
  • Gregorio, Eduard
  • Jiménez-Brenes, Francisco Manuel
  • Sanz, Ricardo
  • Sebé, Francesc
  • Llorens, Jordi
  • Torres-Sánchez, Jorge
© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)., Characterizing crop canopies is especially important in the management of woody crops. In this article, two systems were compared to characterise a 50 m long vineyard row section. One of the systems was a mobile terrestrial laser scanner based on a light detection and ranging (LiDAR) sensor (MTLS-LiDAR). The other was an uncrewed aerial vehicle (UAV) based system using digital aerial photogrammetry (UAV-DAP). The resulting 3D point clouds were assessed qualitatively and quantitatively. Canopy heights, widths and volumes were obtained in 0.1 m long sections along the studied row. All the parameters derived from the two systems presented statistically significant differences. The coefficients of determination between systems were 0.619 for canopy maximum heights above ground level (agl), 0.686 for 90th percentile (P90) heights agl, and 0.283 and 0.274 for maximum and P90 vegetated heights, respectively. Coefficients of determination between averaged maximum canopy width and P90 canopy width were 0.328 and 0.317, respectively. Coefficients of determination between cross-sectional areas determined from maximum widths, P90 widths and from the occupancy grid method were 0.423, 0.409 and 0.334, respectively. Total canopy volume for the entire row obtained from the three cross section estimation methods differed between 19 m3 and 25 m3. The reasons found were that the MTLS-LiDAR-derived point cloud captured the canopy top and side variability but could be affected by occlusions, mixed pixels and tall grass-like weeds present in the surveyed area. For its part, the UAV-DAP-derived point cloud tended to miss top and side shoots and somewhat smoothed canopy variability. As neither of the systems is optimal, a balance needs to be found according to the specific requirements of the survey. For this purpose, a list of pros and cons is presented to support the selection of one of the two systems for canopy monitoring. The MTLS-LiDAR system should be chosen when high detail is required but small areas are to be scanned. Alternatively, the UAV-DAP system should be chosen when large areas are to be monitored and when canopy detail is not so important. Further results are presented in Part 2 for a larger area and including pear and peach orchards with different training systems. Future research is to be conducted on how the compared systems affect variability detection and support variable-rate prescriptions., This research was funded by the projects AGVANCE (AGL2013-48297-C2-2-R), PAgPROTECT (PID2021-126648OB-I00), DECIMAL (PID2020-113229RB-C41) and PRODIGIA (PID2020-113229RB-C44) from the Spanish Ministry of Science and Innovation MICIN/AEI/10.13039/501100011033/FEDER, UE., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Research data Set de datos (Dataset)

Unmanned Aerial Vehicle Imagery for Early Stage Weed Classification and Detection in Maize and Tomato Crops

  • Mesías-Ruiz, Gustavo A.
  • Borra-Serrano, Irene
  • Peña Barragán, José Manuel
  • Castro, Ana Isabel de
  • Fernández-Quintanilla, César
  • Dorado, José
[Description of methods used for collection/generation of data] Data were acquired through an RGB camera mounted on a UAV at an altitude of 11m. From the images obtained by the UAV, geomatic products were systematically generated for the creation of orthomosaics. This task was performed by incorporating information extracted from the RGB channels using Agisoft PhotoScan software (Agisoft LLC, St. Louis, MO). The resulting orthomosaics were subdivided into smaller sections. To identify and label weed species, malherbology experts carried out the task manually. This involved drawing bounding boxes around each plant and annotating various visible objects present in each divided image. This process was executed using the graphical tool labelImg open source software. The final result consists of a collection of individual images, each corresponding to a specific label, i.e., to each plant of each identified species., The dataset, titled 'Unmanned Aerial Vehicle Imagery for Early Stage Weed Classification and Detection in Maize and Tomato Crops,' comprises RGB images captured from an unmanned aerial vehicle flying at an altitude of 11 meters. These images were taken during the early growing season of two summer crops: Maize (BBCH14) at the CSIC experimental farm La Poveda in Arganda del Rey, Madrid, Spain, and Tomato (BBCH501) in commercial plots located in Santa Amalia, Badajoz, Spain. The dataset includes 33,467 labels representing weeds (Atriplex patula, Chenopodium album, Convolvulus arvensis, Cyperus rotundus, Lolium rigidum, Portulaca oleracea, Salsola kali, Solanum nigrum) and crops (maize, tomato). All images are saved in *.jpg format and have been labeled using partitions extracted from an orthomosaic. The final result consists of a collection of individual images, each corresponding to a specific label, i.e., to each plant of each identified species., This work was supported by the Spanish Research State Agency (AEI) through the Project PID2020-113229RB-C41/AEI/10.13039/501100011033. The lead author, G.A. Mesías-Ruiz has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Cognitive computing for classification of six weed species in tomato and maize crops

  • Mesías-Ruiz, Gustavo A.
  • Dorado, José
  • Castro, Ana Isabel de
  • Peña Barragán, José Manuel
ECPA2023 14th European Conference on Precision Agriculture, 2-6 July 2023, Bologna, Italy, Agriculture has used technology as a tool to reinvent itself over time. Cognitive computing is used in numerous artificial intelligence applications, including expert systems, robotics, virtual reality, and machine learning. Supervised deep learning models require a large amount of labeled data for training, representing a common challenge that affects current studies of weed detection and classification using images acquired from a drone. The data augmentation technique provides a practical solution to alleviate insufficient datasets. In this study, we investigate the capability of a cognitive computational system applied to a real-world scenario for data generation, we use the pre-trained Generative Face Priority (GFP) model, which is employed in blind face restoration by means of spatial feature transformation layers, and this model is based on Generative Adversarial Neural Networks (GAN). Data augmentation was applied for weed species found in the early stages of maize and tomato crops; monocots (Cyperus rotundus, Lolium) and dicots (Atriplex patula, Convolvulus arvensis, Salsola kali and Solanum nigrum). The RGB images were acquired from a drone at an altitude of 12 m. The real and augmented datasets were systematically evaluated with those generated by GFP-GAN, for different scaling factors in the images that constitute the input to the classifiers based on the convolutional neural networks VGG16 and ResNet152. Our results show that the technique used improves the generalization capability of the evaluated classifiers, so the size of the dataset directly influences the performance of the models, noticing an approximate increase of more than 5% in the model accuracy metric; furthermore, the GFP-GAN architecture restores with realistic and faithful details by increasing the spatial resolution of the image. The proposed approach demonstrates the scope of the cognitive system built in a domain different from the target domain., This research was financed by the projects PID2020-113229RB C41/AEI/10.13039/501100011033 and PDC2021-121537-C22 (MCIN/AEI/10.13039/501100011033 and NextGenerationEU/PRTR). Research of GAMR was financed by a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

DIGINVASIVE: a digital system to map the presence of invasive weed plants.

  • Castro, Ana Isabel de
  • Mesías-Ruiz, Gustavo A.
  • Llenes-Espigares J.M.
  • Dorado, José
  • Recasens J.
  • Peña Barragán, José Manuel
Invasive plant species are a major cause of crop loss and have high potential to disturb the existing ecological systems [1]. Amarantus palmeri has been recently detected in Mediterranean basin countries, triggering highest alerts of the European Plant Protection Organization [2] and national plant health services, as it is an extremely competitive species native to North America with a high fertility rate and adept at developing herbicide-resistant biotypes [3]. In Spain, A. palmeri has been included among the exotic species that pose a serious threat to productive sectors, pointing to the need for containment efforts and early detection of new cases [4]. In recent years, agricultural web-based tools have made the process of crop monitoring and surveillance easier and more accessible. They are online digital platforms that combine Geographic Information Systems (GPS) technology and big data collected by remote sensing platform or meteorological station, among others, analyze them, and provide graphical and text information in the form of maps and databases. However, they have not been developed for invasive weeds in agricultural domain. Unmanned Aerial Vehicles (UAV) and Artificial Intelligence (AI), as object-based image analysis OBIA and machine learning (ML) procedures, are required to detect and map weeds in challenging scenarios, such as in early season herbaceous crops due to the high spectral similarity of different plant species and the small size of plants in such early stages of development [5,6]. To this end, an open source web-based interface system, called DIGINVASIVE, is being designed to assist end-users by providing an open web mapping and GIS platform to upload, analyze and visualize geo-spatial data that will be able to map the presence of invasive weed plants., This research was partly supported by the projects TED2021-132401A-I00 financied by Next Generation EU, and the Spanish Research State Agency (AEI) through the Project PID2020- 113229RB-C41/AEI/10.13039/501100011033. Research of GMR was financed by a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Otros (Other)

Plant functional diversity is affected by weed management through processes of trait convergence and divergence

  • Guerra, José G.
  • Cabello, Félix
  • Fernández-Quintanilla, César
  • Peña Barragán, José Manuel
  • Dorado, José
© 2022 Guerra, Cabello, Fernández-Quintanilla, Peña and Dorado. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms., Weed management involving tillage and/or herbicides has generally led to a decline of plant diversity in agroecosystems, with negative impacts on ecosystem services provision. The use of plant covers has become the predominant alternative in vineyard management, with numerous studies focusing on analyzing the advantages and disadvantages of plant covers compared to the aforementioned management. Although the impacts of weed management on taxonomic diversity have been widely studied, many gaps remain on their effects on plant functional diversity. As plant functional diversity is linked to the delivery of key ecosystem services in agroecosystems, understanding these effects could enable the development of more sustainable practices. From 2008 to 2018, a long-term trial was carried out in a Mediterranean vineyard to assess different agricultural practices. In this article, we examined how weed management, as well as irrigation use, could affect plant functional diversity. Based on 10 functional traits, such as plant height, specific leaf area or seed mass, we measured different indices of functional diversity and used null models to detect processes of trait convergence and divergence. Our results revealed that weed management and irrigation use had a significant effect on plant functional diversity. Mown plots showed the highest functional richness but were functionally convergent, since mowing was a strong functional filter on most of the traits. Tillage also behaved as a functional filter on some vegetative traits, but favored the divergence of certain reproductive traits. Herbicide-treated and irrigated plots showed the highest values of functional divergence by promoting more competitive species with more divergent trait values. The effect of weed management on these community assembly processes was shaped by the use of irrigation in vineyard rows, leading to functional divergence in those vegetative traits related to resource acquisition and seed mass. These results suggest that greater functional diversity may be associated with the bias caused by higher occurrence of competitive species (e.g. Convolvulus arvensis, Sonchus asper) with contrasting values for certain traits. Therefore, since these species are considered harmful to crops, higher plant functional diversity might not be a desirable indicator in agroecosystems., This work was supported by FEDER/MICINN-AEI, grant number AGL2017-83325-C4-1-R and by MICINN-AEI, grant number PID2020-113229RB-C41/AEI/10.13039/501100011033. The lead author has been granted a predoctoral research fellowship FPI-INIA2016-00035., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Detección y clasificación de malas hierbas mediante drones y redes neuronales profundas: creación de mapas para tratamiento localizado

  • Mesías-Ruiz, Gustavo A.
  • Peña, José
  • Castro, Ana Isabel de
  • Borra-Serrano, Irene
  • Dorado, José
5 Pág., [EN] Accurate detection and identification of weeds are essential in the implementation of Precision Agriculture (PA). In this study, we addressed the detection and classification of weeds in maize and tomato in their early growth stages by integrating images acquired from UAVs and analysis based on advanced convolutional neural networks. Subsequently, the objective of this research was the creation of geo-referenced maps that would allow localized and selective post-emergence management of weed species. Our results indicate that the combination of UAV-captured imagery and deep learning algorithm-based image analysis provides an effective solution for this purpose. The accuracy and efficiency achieved in the identification of weeds in their early stage were promising, both in the set of species present in the crop and in individual species. This advance is of great relevance in the field of PA, as it would allow a more efficient management of weeds by selecting the herbicide according to the type of weed present, reducing the use of herbicidesand ultimately contributing to the sustainability and profitability of agriculture. In addition, the generation of geo-referenced maps facilitates real-time decision making. In summary, this study suggests that the combination of emerging technologies, such as drones and deep neural networks, may be applicable tools in localized weed management in the context of PA., [ES] La detección precisa y la identificación de malas hierbas son esenciales en la implementación de la Agricultura de Precisión (AP). En este estudio, se abordó la detección y la clasificación de malas hierbas en maíz y tomate en sus etapas tempranas de crecimiento mediante la integración de imágenes adquiridas desde drones y análisis basado en avanzadas redes neuronales convolucionales. Posteriormente, el objetivo de esta investigación fue la creación de mapas georreferenciados que permitieran una gestión en post-emergencia localizada y selectiva de las especies de malas hierbas. Nuestros resultados indican que la combinación de imágenes captadas por drones y un análisis de las mismas basado en algoritmos de aprendizaje profundo proporciona una solución efectiva para este propósito. La precisión y eficiencia alcanzadas en la identificación de malas hierbas en su etapa temprana fueron prometedoras, tanto en el conjunto de especies presentes en el cultivo como en especies individuales. Este avance es de gran relevancia en el ámbito de la AP, ya que permitiría una gestión más eficiente de las malas hierbas mediante la selección del herbicida según el tipo de mala hierba presente, reduciendo el uso de herbicidas y, en última instancia, contribuyendo a la sostenibilidad y la rentabilidad de la agricultura. Además, la generación de mapas georreferenciados facilita la toma de decisiones en tiempo real. En resumen, este estudio sugiere que la combinación de tecnologías emergentes, como drones y redes neuronales profundas, pueden ser herramientas aplicables en el manejo localizado de malas hierbas en el contexto de la AP., Este trabajo ha sido financiado por la Unión Europea NextGenerationEU/PRTR y la Agencia Estatal de Investigación (MCIN/AEI/10.13039/501100011033)a través de los proyectos DATI-PRIMA (PCI2021-121932) y SmartWeeding (PID2020-113229RBC41),y de las ayudas FPI (PRE2018-083227) y Juan de la Cierva (FJC2021-047687-1) de los investigado-res Gustavo Mesías-Ruiz e Irene Borra-Serrano,respectivamente, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

How weed management influence plant community composition, taxonomic diversity and crop yield: A long-term study in a Mediterranean vineyard

  • Guerra, José G.
  • Cabello, Félix
  • Fernández-Quintanilla, César
  • Peña, José María
  • Dorado, José
© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/), Agricultural expansion and intensification is a major driver of biodiversity loss. Conventional weed management (e.g. tillage, herbicide) has encouraged the promotion of a few more competitive species over a wide range of arable plants that are now in decline. A reduction in plant diversity would negatively affect the provision of key ecosystem services in agroecosystems. In vineyards, the use of plant covers has emerged as the principal alternative to conventional weed management, providing multiple ecosystem services such as enhance biodiversity, but usually with a decrease in yields in Mediterranean vineyards. This long-term study assessed how weed management (herbicide, mowing, tillage) influenced plant community composition, taxonomic diversity and vineyard yield. Plant surveys and grape yield measurements were carried out from 2015 to 2018 in an experiment established in 2008. Effects on plant community composition were analysed, focusing on two groups: noxious grapevine weeds and terophytic grassland species. In addition, three fundamental components of taxonomic diversity were measured: β-diversity, evenness and species richness. Linear mixed models and generalised linear mixed model were used to examine the response of different variables to weed management. Position (row, inter-row) as well as the indirect effect of adjacent management on the subplots were also included as fixed factors. Furthermore, the percentage of bare soil, related to disturbance degree associated with management, was explored as a potential predictor of taxonomic diversity and vineyard yield. Results indicated that plant community composition was affected by weed management and to a lesser extent by position. Noxious grapevine weeds were more abundant in herbicide-sprayed rows, while terophytic grasslands species showed a higher presence in mown subplots. Weed management had a strong effect on bare soil, which proved to be a good predictor of the variables studied. Indeed, the highest species richness were associated with the lowest percentages of bare soil (mown subplots), while the lowest values were found at high percentages of bare soil (tilled subplots). Herbicide showed intermediate values, with a response dependent on adjacent management. On the other hand, vineyard yields were higher with increased bare soil, although with the consequent loss of species richness. Conversely, mowing in spontaneous plant covers caused a decrease in yield, but still production was within the maximum limit established by the Regulatory Council, while increasing the species richness and a significant presence of therophytic grasslands species. These findings could foster the development of more sustainable management in Mediterranean vineyards., This research was funded by the Spanish Agencia Estatal de Investigación (AEI) and the European Regional Development Fund (ERDF) (Projects AGL2017-83325-C4-1-R and PID2020-113229RB-C41/AEI/10.13039/501100011033). The lead author has been beneficiary of a Spanish Predoctoral Research grant FPI-INIA2016–00035., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

DIGINVASIVE: Sistema de Alerta y Control de especies invasoras. Caso de estudio: Amaranthus palmeri, DIGINVASIVE: Alert and Control System for weed invasive species. Case study: Amaranthus palmeri

  • Castro, Ana Isabel de
  • Mesías-Ruiz, Gustavo A.
  • Llenes-Espigares J.M.
  • Borra-Serrano, Irene
  • Dorado, José
  • Recasens J.
  • Peña Barragán, José Manuel
6 Pág., [ES] La gestión eficiente de las malas hierbas invasoras es fundamental para mantener el equilibrio entre biodiversidad, producción y sostenibilidad agrícola. Esto se debe a que las malas hierbas invasoras pueden causar pérdidas significativas de rendimiento en los cultivos y su control supone un elevado coste en productos fitosanitarios, además de los daños ecológicos que generan. Los avances en robótica, digitalización, inteligencia artificial, etc., permiten la optimización de los recursos empleados (e.g. fitosanitarios) mediante la aplicación de tratamientos localizados y ajustados a las necesidades reales del cultivo siguiendo los principios de la agricultura de precisión. Recientemente se ha confirmado la presencia de la planta invasora Amaranthus palmeri en parcelas agrícolas en España. Esta especie es altamente competitiva, presenta una elevada tasa de fecundidad y la capacidad para desarrollar biotipos resistentes a los herbicidas, convirtiéndola en una de las especies exóticas con mayor amenaza para la agricultura española. Para ello, se ha desarrollado una herramienta de agricultura digital -denominada DIGINVASIVE- para el mapeo y seguimiento de A. palmeri en campos agrícolas, consistentes en un servidor de cartografía digital web de código abierto que utiliza tecnologías digitales innovadoras como imágenes de dron, Sistemas de Información Geográfica, Tecnologías de la Información y la Comunicación y potentes algoritmos de Inteligencia Artificial. Esta herramienta interactiva permitirá a los usuarios establecer las medidas de prevención y erradicación necesarias para abordar el control de A. palmeri en escenarios agrícolas., [EN] Efficient invasive weed management is critical to maintain the balance between biodiversity, production and agricultural sustainability. Invasive plants may cause significant yield losses in crops and their control entails a high cost in pesticide products, in addition to the ecological damage they generate. Advances in robotics, digital agriculture, artificial intelligence, etc, allow the inputs’ optimization (e.g. pesticides) through timely and site-specific treatments tailored to the real crop needs, aligned with the principles of precision agriculture. Amaranthus palmeri has been recently detected in Spanish crop fields. It is an extremely competitive species, with a high fertility rate and skilled at developing herbicide-resistant biotypes, which poses a serious threat to Spanish agriculture. Therefore, a digital agriculture tool -called DIGINVASIVE- is being designed for mapping and monitoring of A. palmeri in crops, which consists of an open-source web-based interface system using innovative digital technologies such as drone imagery, Geographic Information Systems, Information and Communication Technologies and powerful Artificial Intelligence algorithms. This interactive tool will allow users to establish suitable prevention and eradication measures to address the control of A. palmeri in agricultural environments., Este trabajo ha sido financiado por la Unión Europea NextGenerationEU/PRTR y la Agencia Estatal de Investigación (MCIN/AEI/10.13039/501100011033)a través de los proyectos DigInvasive (TED2021-132401A-I00) y SmartWeeding (PID2020-113229RBC41/AEI/10.13039/501100011033), y de las ayudas FPI (PRE2018-083227) y Juan de la Cierva (FJC2021-047687-1) de los investigadores GMR e IBS, respectivamente. También agradecemos a David Campos y José Manuel Martín, miembros de Tec4AGRO, por su apoyo en los trabajos de campo y gabinete., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Impacto del manejo localizado en el control de Sorghum halepense y el rendimiento del maíz

  • Peña Barragán, José Manuel
  • Borra-Serrano, Irene
  • Fernández-Quintanilla, César
  • Castro, Ana Isabel de
  • Mesías-Ruiz, Gustavo A.
  • Guerra, José G.
5 Pág., [EN] Sorghum halepense (Johnsongrass) is a weed that tends to spatially distribute in an aggregated manner in maize crops. In a two-year study conducted at La Poveda (CSIC, Madrid), the effectiveness of Precision Agriculture (PA) in controlling this weed was investigated by: 1) monitoring the spatio-temporal dynamics of Johnsongrass patches in a maize field using UAV aerial image processing; 2) site-specific application of the herbicide nicosulfuron according to the early-stage map generated from aerial imagery; and 3) quantification of crop yield in Johnsongrass treated and uuntreated areas, as well as weed-free. The analysis of the map prior to herbicide treatment in 2022 revealed a random distribution of Johnsongrass, with patches averaging 0.4 m² in size, ranging from 0.01 m² to 5 m². Site-specifically treated plots showed spatial growth control, evidenced in the subsequent season (2023) with an average patch size of 1.0 m2, ranging from 0.02 m² to 12 m². In contrast, untreated patches significantly increased in size (average 3.8 m2; ranging between 0.01 m² and 122 m²). There was a notable decrease in yield in untreated infested areas compared to treated and/or Johnsongrass-free areas. These findings underscore the advantages of PA in Johnsongrass control, demonstrating how detection and site-specific patch management can reduce herbicide use by up to 95% and mitigate competition with the maize crop., [ES] Sorghum halepense (cañota) es una mala hierba que tiende a distribuirse espacialmente de forma agregada en los cultivos de maíz. En un estudio de dos años realizado en La Poveda (CSIC, Madrid), se investigó la eficacia de la Agricultura de Precisión (AP) en su control, mediante: 1) el seguimiento de la dinámica espacio-temporal de los rodales de cañota en un campo de maíz con el procesamiento de imágenes aéreas UAV; 2) la aplicación localizada del herbicida nicosulfuron según el mapa generado en etapa temprana con imágenes aéreas; y 3) la cuantificación del rendimiento del cultivo en áreas tratadas y no tratadas con cañota, así como libres de mala hierba. El análisis del mapa previo al tratamiento herbicida en 2022 reveló una distribución aleatoria de cañota, con rodales de tamaño medio de 0,4 m2, variando entre 0,01 m² y 5 m². Las parcelas tratadas con herbicida localizado mostraron un control del crecimiento espacial, evidenciado en la campaña siguiente (2023) con un tamaño promedio de 1,0 m2, oscilando entre 0,02 m² y 12 m². En contraste, los rodales no tratados aumentaron significativamente su tamaño (media 3,8 m2; rango entre 0,01 m² y 122 m²). El rendimiento disminuyó notablemente en zonas infestadas sin tratamiento respecto a las zonas tratadas y/o libres de cañota. Estos resultados subrayan las ventajas de la AP en el control de la cañota, demostrando cómo la detección y manejo localizado de rodales podrían reducir hasta un 95% el uso de herbicidas y mitigar la competencia con el cultivo de maíz., Trabajo financiado por el proyecto PID2020-113229RBC41/AEI/10.13039/501100011033. La ayuda FJC2021-047687-1 a Irene Borra-Serrano, dentro del programa Juan de la Cierva, está financiada por MCIN/AEI/10.13039/501100011033 y fondos de la Unión Europea NextGenerationEU/PRTR., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Weed species classification with UAV imagery and standard CNN models: Assessing the frontiers of training and inference phases

  • Mesías-Ruiz, Gustavo A.
  • Borra-Serrano, Irene
  • Peña Barragán, José Manuel
  • Castro, Ana Isabel de
  • Fernández-Quintanilla, César
  • Dorado, José
12 Pág.-- © 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)., Accurate weed species identification is crucial for effective site-specific weed management (SSWM), enabling targeted and timely control measures for each weed in crop field. This study advanced the current approach to species-level weed identification during the early growth stage by integrating unmanned aerial vehicles (UAVs) imagery with standard convolutional neural networks (CNNs) models such as VGG16, Resnet152 and Inception-Resnet-v2. For this, a robust dataset was created with 33,467 labels of weeds (Atriplex patula, Chenopodium album, Convolvulus arvensis, Cyperus rotundus, Lolium rigidum, Portulaca oleracea, Salsola kali, Solanum nigrum) and crops (maize, tomato), which was subjected to different training, validation and test scenarios. Model inputs were adjusted in order to align them with the information represented by the UAV images. Initially, models were developed in balanced scenarios, gradually increasing label numbers to assess their performance. Inception-ResNet-v2 achieved over 90% accuracy with 400 labels, while ResNet152 and VGG16 required 600 and 800 labels, respectively, for similar accuracy. In a more complex and realistic scenarios with unbalanced datasets, Inception-ResNet-v2 outperformed, likely due to its deeper architecture and enhanced capability to capture intricate features and patterns within UAV images. The study emphasized the importance of the minority-to-majority species ratio in unbalanced datasets, which affects minority species classification. To prevent misclassification, it is crucial to determine the right number of labels for CNN model training and validation. Weed maps were generated after species classification using the Faster R–CNN algorithm as an object detector. This advancement in methodology facilitates the precise and efficient implementation of SSWM techniques., This work was supported by the Spanish Research State Agency (AEI) through the Project PID2020-113229RB-C41/AEI/10.13039/501100011033. The lead author, G.A. Mesías-Ruiz has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227). The research of I. Borra-Serrano was financed by the grant FJC2021-047687-1 funded by MCIN/AEI/10.13039/501100011033 and European Union NextGenerationEU/PRTR., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Detección de crucíferas en márgenes de biodiversidad en ecosistemas agrarios mediante el análisis de imágenes dron y arquitecturas CNN, Detection of crucifers in biodiversity margins of agricultural ecosystems through the analysis of drone images and architectures CNN

  • Mena, Juan Diego
  • Dorado, José
  • Guerra, José G.
  • Mesías-Ruiz, Gustavo A.
  • Peña Barragán, José Manuel
  • Borra-Serrano, Irene
[ES] La pérdida generalizada de biodiversidad en los paisajes y ecosistemas agrarios es una preocupación a nivel global. Esta pérdida se relaciona con la intensificación de las prácticas agrícolas, el uso inadecuado de agroquímicos y la fragmentación de hábitats naturales. Todo esto afecta la diversidad vegetal y tiene un impacto en el funcionamiento del ecosistema, limitando los servicios ecosistémicos. Promover una mayor diversidad funcional en los márgenes de nuestros cultivos es una meta alcanzable mediante el manejo de los recursos que nos ofrece la flora arvense. En particular, las especies de la familia Brassicaceae Burnett (en adelante, crucíferas), desempeñan un papel fundamental en los agrosistemas al atraer enemigos naturales de plagas y proporcionar néctar y polen a los polinizadores. El objetivo de este estudio era detectar y estimar los recursos florales ofrecidos por las crucíferas en presencia de otras especies arvenses. Para ello, se diseñó un experimento en la finca “La Poveda” (CSIC, Madrid) en 2023, en el que se llevaron a cabo hasta 4 vuelos con un dron sobre los márgenes florales donde se encontraban tres crucíferas de interés: Diplotaxis tenuifolia, Eruca vesicaria y Moricandia arvensis. Estas especies representan algunos de los rasgos funcionales clave de las crucíferas que suelen estar presentes en nuestros agrosistemas. La detección de estas especies se realizó mediante el análisis de imágenes capturadas por el dron y la implementación de arquitecturas de redes neuronales convolucionales. En este trabajo, presentamos los primeros resultados de los modelos que estamos desarrollando., [EN] The widespread loss of biodiversity in agricultural landscapes and ecosystems is a global concern. This loss is related to the intensification of agricultural practices, the inappropriate use of agrochemicals and the fragmentation of natural habitats. All of this affects plant diversity and has an impact on the functioning of the ecosystem, limiting ecosystem services. Promoting greater functional diversity in the margins of our crops is an achievable goal through the management of the resources offered by the weed flora. In particular, species of the Brassicaceae Burnett family (hereinafter referred to as cruciferous), play a fundamental role in agrosystems by attracting natural enemies of pests and providing nectar and pollen to pollinators. The objective of this study was to detect and estimate the floral resources offered by cruciferous plants in the presence of other weed species. To this end, an experiment was designed at the “La Poveda” farm (CSIC, Madrid) in 2023, in which up to 4 flights were carried out with a drone over the floral margins where three cruciferous plants of interest were found: Diplotaxis tenuifolia, Eruca vesicaria and Moricandia arvensis. These species represent some of the key functional traits of crucifers that are often present in our agrosystems. The detection of these species was carried out through the analysis of images captured by the drone and the implementation of convolutional neural network architectures. In this work, we present the first results of the models we are developing., Al Ministerio de Ciencia e Innovación del Gobierno de España MCIN/AEI/10.13039/501100011033, que financia las ayudas PRE2018-083227 (Gustavo Mesías-Ruiz) y PRE2021-101068 (Juan Diego Mena Castillo) dentro del programa FPI, y al proyecto PID2020-113229RBC41/AEI/10.13039/501100011033. La ayuda FJC2021-047687-1 a Irene Borra-Serrano, dentro del programa Juan de la Cierva, está financiada por MCIN/AEI/10.13039/501100011033 y fondos de la Unión Europea NextGenerationEU/PRTR., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

El uso de drones en el manejo agroecológico de las malas hierbas, Using drones for agroecological weed management

  • Fernández-Quintanilla, César
  • Dorado, José
  • Peña Barragán, José Manuel
[EN] Drones are a valuable tool to explore the spatial variability present in crop fields. Although its use is becoming relatively common in technological oriented systems it is very rare in ecological oriented systems. The goal of this presentation is to explore some of the potentialities of this tool to contribute to an improved agroecological weed management. The information obtained in early flights can be used to minimize input use (tillage operations or herbicides). The information obtained at the end of the crop life cycle can be used to assess the efficacy of the control measures, monitoring weed escapes and/or resistant populations. Digital phenotyping from aerial images could be used in breeding weed suppressing varieties. The assessment of agronomic practices under real conditions can be conducted through aerial monitoring farmer`s plots with variable conditions. The large volume of information (big data) provided on soil, crop and weed conditions will facilitate the design and implementation of integrated weed management systems. The large volume of information (big data) provided on soil, crop and weed conditions will facilitate the design and implementation of integrated weed management systems. The use of drones at the landscape scale allow to monitor biodiversity spaces and to assess the ecosystem services provided by these areas., [ES] Los drones constituyen una herramienta muy valiosa para explorar la variabilidad espacial presente en las parcelas de cultivo. Sin embargo, su uso, cada vez más frecuente en sistemas con una orientación tecnológica, es prácticamente desconocida su aplicación dentro de sistemas para una aproximación agro-ecológica. El objetivo de este trabajo es explorar algunas de las potencialidades de los drones para contribuir a un mejor manejo agroecológico de las malas hierbas. La información extraída con vuelos tempranos nos permite minimizar el uso de insumos (labores, herbicidas) y mejorar su eficiencia. Por otra parte, la información obtenida en vuelos realizados al final del ciclo nos permite evaluar la eficacia de las medidas tomadas, detectando la aparición de escapes o de poblaciones resistentes. El fenotipado digital a partir de imágenes de dron puede ser de gran utilidad en los programas de mejora de variedades de cultivos supresoras de malas hierbas. La evaluación de prácticas agronómicas en condiciones reales puede llevarse a cabo mediante la monitorización continuada de parcelas de agricultores en las que exista una diversidad de condiciones. La abundante información suministrada por los drones sobre el estado del suelo, cultivo y poblaciones de plagas y malas hierbas facilitaría el diseño y la implementación de sistemas de gestión integrada. El empleo de drones a escala paisaje nos permite el seguimiento de los espacios de biodiversidad y evaluar los servicios ecosistémicos obtenidos., Trabajo financiado por el proyecto PID2020-113229RBC41/AEI/10.13039/501100011033., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Transfer and zero-shot learning for scalable weed detection and classification in UAV images

  • Belissent, Nicolas
  • Peña Barragán, José Manuel
  • Mesías-Ruiz, Gustavo A.
  • Shawe-Taylor, John
  • Pérez-Ortiz, María
© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/), In an effort to reduce pesticide use, agronomists and computer scientists have joined forces to develop site-specific weed detection and classification systems. These systems aim to recognize and locate weed species within a crop field, using precision equipment to apply required herbicides timely and only where needed, with the objective of reducing the sprayable surface required to eliminate the given weed and protect the crop, with both economic and environmental benefits. Yet, with climate change on the rise, common weeds are expected to undergo some changes to adapt to their environment, possibly with new or invasive weeds spreading to areas where they did not exist before. These changes (often morphological) as well as new invasions need to be taken into account by future classifiers and detection algorithms to ensure system robustness and adaptation to new habitats/climate dynamics. This paper proposes a set of experiments evaluating the use of transfer learning and zero-shot learning for weed classification using our novel TomatoWeeds dataset. Residual networks of variable depth, pretrained on the Imagenet and/or DeepWeeds datasets were evaluated. A ResNet50 pretrained on both datasets and fine-tuned on the TomatoWeeds dataset performed best, returning a holdout set accuracy of 77.8%, showing the advantageous use of transfer learning in this domain. Zero-shot learning, using both embeddings of images and morphological and habitat text-based descriptions, is implemented to test the ability of machine learning pipelines of recognizing unseen classes at test time (which may arise e.g. due to changing climate dynamics), a learning task in which the field (and our experiments) are still far from satisfactory results. Further research could benefit from larger weed-specific datasets for transfer learning as well as deeper network architectures to improve model performance. The projection-based ZSL could also benefit from larger datasets and new zero-shot learning architectures in hope that unseen classes are accurately projected., This work was partially supported by the Spanish Research State Agency (AEI/10.13039/501100011033) and the NextGenerationEU/PRTR through the Projects PDC2021-121537-C22 and PID2020-113229RB-C41. The author GM-R has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Research data Set de datos (Dataset)

DRONEWEED: DRONE imagery dataset for early-season WEED classification

  • Mesías-Ruiz, Gustavo A.
  • Peña Barragán, José Manuel
  • Castro, Ana Isabel de
  • Dorado, José
[Description of methods used for collection/generation of data] Data were acquired through an RGB camera mounted on a UAV at an altitude of 11m., Identification of weed species at the early stages of growth is critical for precision agriculture. Accurate detection and classification to species level allows targeted control measures to be taken, significantly reducing pesticide use. This dataset consists of RGB images, captured with a Sony ILCE-6300L camera mounted on an unmanned aerial vehicle (UAV) at 11 meters altitude. The dataset covers several agricultural fields in Spain, focusing on two summer crops: corn and tomato. It is designed to improve the accuracy of early season weed detection by including images from two phenological stages. Specifically, the dataset contains 31,002 labeled images from the early growth stage - maize with four leaves unfolded (BBCH14) and tomato with the first flower bud visible (BBCH501) - as well as 36,556 images from a later growth stage - maize with seven leaves unfolded (BBCH17) and tomato with the ninth flower bud visible (BBCH509). In maize, weed species include Atriplex patula, Chenopodium album, Convolvulus arvensis, Datura ferox, Lolium rigidum, Salsola kali and Sorghum halepense. In tomato, weed species include Cyperus rotundus, Portulaca oleracea and Solanum nigrum. The images, stored in JPG format, were labeled by partitioning orthomosaics, with each image corresponding to a specific plant species. This dataset is ideal for developing advanced deep learning models, such as CNN and ViT, for early detection and classification of weed species in corn and tomato crops using UAV imagery. By providing this dataset, we aim to advance UAV-based weed detection and mapping technologies, contributing to precision agriculture with more efficient and accurate tools that promote sustainable and profitable agricultural practices., This work was supported by the Spanish Research State Agency (AEI) through the Project PID2020-113229RB-C41/AEI/10.13039/501100011033. The lead author, G.A. Mesías-Ruiz has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Cognitive Computing Advancements: Improving Precision Crop Protection through UAV Imagery for Targeted Weed Monitoring

  • Mesías-Ruiz, Gustavo A.
  • Peña Barragán, José Manuel
  • Castro, Ana Isabel de
  • Borra-Serrano, Irene
  • Dorado, José
20 Pág., Early detection of weeds is crucial to manage weeds effectively, support decision-making and prevent potential crop losses. This research presents an innovative approach to develop a specialized cognitive system for classifying and detecting early-stage weeds at the species level. The primary objective was to create an automated multiclass discrimination system using cognitive computing, regardless of the weed growth stage. Initially, the model was trained and tested on a dataset of 31,002 UAV images, including ten weed species manually identified by experts at the early phenological stages of maize (BBCH14) and tomato (BBCH501). The images were captured at 11 m above ground level. This resulted in a classification accuracy exceeding 99.1% using the vision transformer Swin-T model. Subsequently, generative modeling was employed for data augmentation, resulting in new classification models based on the Swin-T architecture. These models were evaluated on an unbalanced dataset of 36,556 UAV images captured at later phenological stages (maize BBCH17 and tomato BBCH509), achieving a weighted average F1-score ranging from 94.8% to 95.3%. This performance highlights the system’s adaptability to morphological variations and its robustness in diverse crop scenarios, suggesting that the system can be effectively implemented in real agricultural scenarios, significantly reducing the time and resources required for weed identification. The proposed data augmentation technique also proved to be effective in implementing the detection transformer architecture, significantly improving the generalization capability and enabling accurate detection of weeds at different growth stages. The research represents a significant advancement in weed monitoring across phenological stages, with potential applications in precision agriculture and sustainable crop management. Furthermore, the methodology showcases the versatility of the latest generation models for application in other knowledge domains, facilitating time-efficient model development. Future research could investigate the applicability of the model in different geographical regions and with different types of crops, as well as real-time implementation for continuous field monitoring., This research was funded by the SPANISH RESEARCH STATE AGENCY (AEI) through the project PID2020-113229RB-C41/AEI/10.13039/501100011033. The lead author, G.A. Mesías-Ruiz has been a beneficiary of an FPI fellowship by the SPANISH MINISTRY OF EDUCATION AND PROFESSIONAL TRAINING, grant number PRE2018-083227-. The research of I. Borra-Serrano was financed by the grant FJC2021-047687-1 funded by MCIN/AEI/10.13039/501100011033 and EUROPEAN UNION NEXTGENERATIONEU/PRTR., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Quantifying the Compatibility of Optical Reflectance Factors in a Field Intercomparison Experiment

  • Pacheco-Labrador, Javier
  • Peón, Juanjo
  • Jiménez, Marcos
  • Rodríguez-Pérez, José Ramón
  • Jiménez-Berni, José A.
  • Aragonés, David
  • Díaz-Delgado, Ricardo
  • Dorado, José
  • Castro, Ana Isabel de
  • Martín, M. Pilar
This work presents the results of an outdoor intercomparison experiment where the reflectance factors from five full-range field spectroradiometers (400–2500 nm) were compared using repeatability (RPT), reproducibility (RPR), and compatibility metrics. We confronted four analytical spectral devices (ASDs) and one Spectravista Corporation (SVC) sensor. Ancillary atmospheric characterization proved useful for filtering data acquired under suboptimal sky conditions. RPT uncertainties were low, while RPR uncertainty was significantly larger due to changes in the measured surface between measurements and directional effects induced by varying sun angles and illumination conditions. Compatibility proved to be a valuable metric for interpreting the intercomparison results and supports the importance and necessity for uncertainty quantification in field spectroradiometry. RPR uncertainty in dark heterogeneous surfaces, such as vegetation, might limit the identification of subtle differences associated with their biochemistry and require specific protocols that minimize these errors., This work was supported in part by Spanish National Research Council (CSIC) Interdisciplinary Thematic Platform (PTI) under Grant PTI-TELEDETECT; in part by Spanish Network of Optical Proximal Sensing (NetOPS) through MICIU/AEI/10.13039/501100011033 under Grant RED2022-134438-T; in part by Spanish Ministry of Science, Innovation and Universities through MCIN/AEI/10.13039/501100011033 under Project TED2021-129814B-I00; in part by European Union (EU) NextGenerationEU/PRTR; in part by the Sustainability for Mediterranean Hotspots in Andalusia, under Grant LIFEWATCH-2019-09-CSIC-4, POPE 2014–2020; in part by the SpaFLEXVal under Grant PID2022-137022OB-C33; in part by Fluorescence Explorer FLEX-S3 under Grant PCI2023-145988-2; in part by the eLTER Plus Project (INFRAIA- Integrating Activities for Starting Communities, Horizon 2020) under Agreement 871128; in part by European Union Recovery, Transformation and Resilience Funds, and NextGenerationEU/PRTR; and in part by the Partnership on Research and Innovation in the Mediterranean Area (PRIMA EU) through the Digital Agriculture Technologies for Irrigation efficiency (DATI)-Project “PCI2021-121932” and DECIMAL-Project “PID2020-113229RB-C41.” The work of Ricardo Díaz-Delgado was supported by “Salvador de Madariaga” through Spanish Ministry of Science, Innovation and Universities under Grant PRX22/00726, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

A comparison of environmental impacts of three contrasting cropping systems for barley production under Mediterranean conditions

  • Fernández-Quintanilla, César
  • Lechón, Yolanda
  • Lago, Carmen
  • Peña, José Manuel
  • Dorado, José
Agriculture is a key contributor to environmental degradation and to global change. Consequently, the design of sustainable agricultural systems and the assessment of their relevance is a major priority for European agriculture. Different cropping systems, with variable objectives and constraints, can be used in cereal production in Spain. This study focused in comparing three winter barley cropping systems, ranging from intensive no-till to organic approaches. To assess the environmental impacts of each system, a Life Cycle Assessment was conducted. The findings indicate that the impacts varied depending on the chosen functional unit. When land area was considered the functional unit, the lowest impacts were obtained in the organic system, while the no-till system showed the highest. This difference was primarily attributed to variations in N fertilization. Nitrogen use had a significant impact across various categories, primarily due to the energy demands for its production and transportation, as well as the emissions of NH3 and N2O. However, when evaluating agricultural goods production as the functional unit, the organic system exhibited the highest impacts in terms of energy demand, freshwater ecotoxicity and freshwater eutrophication. These differences are explained by the loss of production in the fallow year and the low yields of the legume crop. The middle-way option provided the lowest impacts when economic net revenues were considered. The main reason for this was its higher total revenues associated to high crop production and EU subsidies., This work was supported by the Spanish Research State Agency (AEI) through the Project PID2020–113229RB-C41/AEI/10.13039/501100011033., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Research data Set de datos (Dataset)

Drone imagery dataset for early-season weed classification in maize and tomato crops, Drone imagery dataset for early-season weed detection and classification in maize and tomato crops

  • Mesías-Ruiz, Gustavo A.
  • Peña Barragán, José Manuel
  • Castro, Ana Isabel de
  • Dorado, José
Identifying weed species at early-growth stages is critical for precision agriculture. Accurate classification at the species-level enables targeted control measures, significantly reducing pesticide use. This paper presents a dataset of RGB images captured with a Sony ILCE-6300L camera mounted on an unmanned aerial vehicle (UAV) flying at an altitude of 11 m above ground level. The dataset covers various agricultural fields in Spain, focusing on two summer crops: maize and tomato. It is designed to enhance early-season weed classification accuracy by including images from two phenological stages. Specifically, the dataset contains 31,002 labeled images from the early-growth stage-maize with four unfolded leaves (BBCH14) and tomato with the first flower bud visible (BBCH501)-as well as 36,556 images from a more advanced-growth stage-maize with seven unfolded leaves (BBCH17) and tomato with the ninth flower bud visible (BBCH509). In maize, the weed species include Atriplex patula, Chenopodium album, Convolvulus arvensis, Datura ferox, Lolium rigidum, Salsola kali and Sorghum halepense. In tomato, the weed species include Cyperus rotundus, Portulaca oleracea and Solanum nigrum. The images, stored in JPG format, were labeled in orthomosaic partitions, with each image corresponding to a specific plant species. This dataset is ideally suited for developing advanced deep learning models, such as CNNs and ViTs, for early classification of weed species in maize and tomato crops using UAV imagery. By providing this dataset, we aim to advance UAV-based weed detection and mapping technologies, contributing to precision agriculture with more efficient, accurate tools that promote sustainable and profitable farming practices., This research was funded by the projects PID2020-113229RB-C41, PID2023-150108OB-C31 and PID2023-150108OB-C32 from the Spanish MCIU/AEI/10.13039/501100011033 and FEDER, EU. The lead author, G.A. Mesías-Ruiz has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Control de malas hierbas en maíz en el contexto de la agricultura de precisión

  • Castro, Ana Isabel de
  • Martín, Carlos
  • Montull, José María
  • Peña Barragán, José Manuel
La gestión de los cultivos evoluciona con los tiempos adaptándose a las nuevas herramientas disponibles. En el caso de las malas hierbas, las prácticas de manejo sehan visto condicionadas también por aspectos legales, ambientales y socio-económicos. Esta situación se refleja claramente en el cultivo del maíz, de alta importancia tanto a nivel mundial como nacional, cuyas producciones son gravemente afectadas por las infestaciones de malas hierbas y su control se ha visto mermado por las nuevas normativas. Sin embargo, las recientes herramientas tecnológicas y digitales generan nuevas oportunidades para el control óptimo de las malas hierbas en maíz en el contexto de la Agricultura de Precisión, con aplicaciones localizadas, dosis adecuadas y en el momento oportuno., Investigación parcialmente financiada por los proyectos DECIMAL (PID2020-113229RB-C41), UAV4LessPest (PDC2021-121537-C22) y DigInvasive (TED2021-132401A-I00) de la Agencia Estatal de Investigación (MCIN/ AEI/10.13039/501100011033), Fondos de Recuperación, Transformación y Resiliencia, y Unión Europea NextGenerationEU/PRTR. Agradecemos también la participación de Bayer C.S. y sus distribuidores Sertagro S.L., Bardenas Comercial S.L. y Fertiplan/Agrinova., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Hacia un 50% menos de herbicida en maíz combinando drones, el sistema IPMwise y pulverizadores comerciales de dosis variable

  • Peña Barragán, José Manuel
  • Montull, José María
  • Martín, Carlos
  • Castro, Ana Isabel de
Se describen unos ensayos piloto llevados a cabo para evaluar la aplicación localizada y variable de herbicidas con dos pulverizadores comerciales dotados de barra de aplicación boquilla a boquilla, siguiendo los principios de la agricultura de precisión. Los ensayos se realizaron en cuatro parcelas comerciales de maíz en post-emergencia temprana, en donde se generaron mapas de prescripción de herbicida a partir de imágenes tomadas con drones y el sistema de soporte a la decisión IPMwise. Dichos mapas dirigieron correctamente la operación de los pulverizadores, obteniéndose un ahorro de herbicidas entre el 38% y 55% en comparación con aplicar un tratamiento convencional, siendo un resultado alineado con los objetivos agroambientales marcados por la Unión Europea sobre reducción en el uso de fitosanitarios., Este trabajo ha sido fruto de una colaboración promovida por BayerCropscience SL, con la participación de las empresas IPMadvice SL, Bardenas Comercial SL, Setagro SL y Agrinova, e investigadores del ICA-CSIC y el INIA-CSIC. Investigación parcialmente financiada por los proyectos DECIMAL (PID2020-113229RB-C41), UAV4LessPest (PDC2021-121537-C22) y DigInvasive (TED2021-132401A-I00) de la Agencia Estatal de Investigación (MCIN/ AEI/10.13039/501100011033), Fondos de Recuperación, Transformación y Resiliencia, y Unión Europea NextGenerationEU/PRTR., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Otros (Other)

Boosting precision crop protection towards agriculture 5.0 via machine learning and emerging technologies: A contextual review

  • Mesías-Ruiz, Gustavo A.
  • Pérez-Ortiz, María
  • Dorado, José
  • Castro, Ana Isabel de
  • Peña Barragán, José Manuel
Crop protection is a key activity for the sustainability and feasibility of agriculture in a current context of climate change, which is causing the destabilization of agricultural practices and an increase in the incidence of current or invasive pests, and a growing world population that requires guaranteeing the food supply chain and ensuring food security. In view of these events, this article provides a contextual review in six sections on the role of artificial intelligence (AI), machine learning (ML) and other emerging technologies to solve current and future challenges of crop protection. Over time, crop protection has progressed from a primitive agriculture 1.0 (Ag1.0) through various technological developments to reach a level of maturity closelyin line with Ag5.0 (section 1), which is characterized by successfully leveraging ML capacity and modern agricultural devices and machines that perceive, analyze and actuate following the main stages of precision crop protection (section 2). Section 3 presents a taxonomy of ML algorithms that support the development and implementation of precision crop protection, while section 4 analyses the scientific impact of ML on the basis of an extensive bibliometric study of >120 algorithms, outlining the most widely used ML and deep learning (DL) techniques currently applied in relevant case studies on the detection and control of crop diseases, weeds and plagues. Section 5 describes 39 emerging technologies in the fields of smart sensors and other advanced hardware devices, telecommunications, proximal and remote sensing, and AI-based robotics that will foreseeably lead the next generation of perception-based, decision-making and actuation systems for digitized, smart and real-time crop protection in a realistic Ag5.0. Finally, section 6 highlights the main conclusions and final remarks., This work was supported by the Spanish Research State Agency (AEI) through the Projects PDC2021-121537-C22/AEI/10.13039/501100011033 and PID2020-113229RB-C41. The lead author GM-R has been a beneficiary of a FPI fellowship by the Spanish Ministry of Education and Professional Training (PRE2018-083227)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Artículo científico (JournalArticle)

Weed control efficacy and economic profitability of three contrasting cropping systems in semiarid rainfed zones

  • Dorado, José
  • Barroso, Judit
  • Peña, José María
  • Luna, Ignacio M.
  • Fernández-Quintanilla, César
Promoting sustainable agriculture in the semiarid, rainfed areas of central Spain requires understanding how weed populations respond to different cropping systems and evaluating the economic profitability of these systems. A 6-yr field experiment compared three systems: a no-till barley (Hordeum vulgare L.) monoculture with fertilizers and herbicides (NT); a 2-yr rotation of grain legume or false flax [Camelina sativa (L.) Crantz] followed by barley, with reduced tillage and agrochemicals (MW); and a 3-yr fallow–grain legume–barley rotation under organic practices (ORG). Results showed that NT consistently provided the most effective weed control, reducing total weed density by 79% to 84% compared with ORG over the last 3 yr of the study. MW also significantly reduced total weed density by 11% to 75% relative to ORG, although some species increased. ORG was the least effective system, with weed densities reaching 395 plants m−2 and the lowest control efficacy. Weed population dynamics varied by management. Rigid ryegrass (Lolium rigidum Gaudin) was effectively controlled in NT through a consistent annual herbicide program and in MW by combining periodic tillage with targeted herbicide applications; however, it was extremely difficult to manage in ORG. Corn poppy (Papaver rhoeas L.) increased over time in all systems, while other species showed no clear trends. Weed species diversity was lowest in NT, with mean species richness reduced by 34% to 39% compared with ORG and by 13% to 17% compared with MW. Economic analysis revealed the highest adjusted net returns in MW and the lowest in ORG, driven by differences in crop sequences, yields, and subsidies from the European Common Agricultural Policy. These findings offer valuable insights for optimizing weed management in rainfed cereal systems, highlighting the importance of integrated approaches that balance effective weed control, crop productivity, and economic viability., Research funded by the projects PID2020-113229RB-C41 and PID2023-150108OB-C31 from the Spanish MCIU/AEI/10.13039/501100011033 and FEDER, EU. We gratefully acknowledge grant PRX22/00630 awarded to JD as part of the Spanish Mobility Subprogram funded by MCIU within the PEICTI 2021-2023 framework., Peer reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters - Part 1: Methodology and comparison in vineyards

  • Escolà i Agustí, Alexandre
  • Peña, José M.
  • López-Granados, Francisca
  • Rosell Polo, Joan Ramon
  • de Castro, Ana I.
  • Gregorio López, Eduard
  • Jiménez-Brenes, Francisco M.
  • Sanz Cortiella, Ricardo
  • Sebé Feixas, Francesc
  • Llorens Calveras, Jordi
  • Torres-Sánchez, Jorge
Characterizing crop canopies is especially important in the management of woody crops. In this article, two systems were compared to characterise a 50 m long vineyard row section. One of the systems was a mobile terrestrial laser scanner based on a light detection and ranging (LiDAR) sensor (MTLS-LiDAR). The other was an uncrewed aerial vehicle (UAV) based system using digital aerial photogrammetry (UAV-DAP). The resulting 3D point clouds were assessed qualitatively and quantitatively. Canopy heights, widths and volumes were obtained in 0.1 m long sections along the studied row. All the parameters derived from the two systems presented statistically significant differences. The coefficients of determination between systems were 0.619 for canopy maximum heights above ground level (agl), 0.686 for 90th percentile (P90) heights agl, and 0.283 and 0.274 for maximum and P90 vegetated heights, respectively. Coefficients of determination between averaged maximum canopy width and P90 canopy width were 0.328 and 0.317, respectively. Coefficients of determination between cross-sectional areas determined from maximum widths, P90 widths and from the occupancy grid method were 0.423, 0.409 and 0.334, respectively. Total canopy volume for the entire row obtained from the three cross section estimation methods differed between 19 m3 and 25 m3. The reasons found were that the MTLS-LiDAR-derived point cloud captured the canopy top and side variability but could be affected by occlusions, mixed pixels and tall grass-like weeds present in the surveyed area. For its part, the UAV-DAP-derived point cloud tended to miss top and side shoots and somewhat smoothed canopy variability. As neither of the systems is optimal, a balance needs to be found according to the specific requirements of the survey. For this purpose, a list of pros and cons is presented to support the selection of one of the two systems for canopy monitoring. The MTLS-LiDAR system should be chosen when high detail is required but small areas are to be scanned. Alternatively, the UAV-DAP system should be chosen when large areas are to be monitored and when canopy detail is not so important. Further results are presented in Part 2 for a larger area and including pear and peach orchards with different training systems. Future research is to be conducted on how the compared systems affect variability detection and support variable-rate prescriptions., This research was funded by the projects AGVANCE (AGL2013-48297-C2-2-R), PAgPROTECT (PID2021-126648OB-I00), DECIMAL (PID2020-113229RB-C41) and PRODIGIA (PID2020-113229RB-C44) from the Spanish Ministry of Science and Innovation MICIN/AEI/10.13039/501100011033/FEDER, UE. The authors also want to thank Codorníu SA and Celler Raimat for allowing the use of one of their vineyards to conduct the trial described in this work.
Recercat. Dipósit de la Recerca de Catalunya
Publication Artículo científico (JournalArticle)

Mobile terrestrial laser scanner vs. UAV photogrammetry to estimate woody crop canopy parameters - Part 2: Comparison for different crops and training systems

  • Torres-Sánchez, Jorge
  • Escolà i Agustí, Alexandre
  • de Castro, Ana I.
  • López-Granados, Francisca
  • Rosell Polo, Joan Ramon
  • Sebé Feixas, Francesc
  • Jiménez-Brenes, Francisco M.
  • Sanz Cortiella, Ricardo
  • Gregorio López, Eduard
  • Peña, José M.
The measurement of geometric canopy parameters in woody crops is an important task in Precision Agriculture because of their correlation with crop condition and productivity. In recent years, several technological approaches have been developed as an alternative to manual measurements, which are time- and labour-consuming. Two of the most commonly used 3D canopy characterization technologies are mobile terrestrial laser scanning (MTLS) based on light detection and ranging (LiDAR) sensors, and digital aerial photogrammetry (DAP) using imagery from uncrewed aerial vehicles (UAVs). Although both are state-of-the-art and have been fully tested and validated, a complete comparison between their geometric canopy parameter estimations in different woody crops and training systems has not been carried out. For this reason, a set of geometric parameters (canopy height, projected area, and volume) of a vineyard, an intensive peach orchard, and an intensive pear orchard were measured using UAV-DAP and MTLS-LiDAR. A comparison between both kinds of measurements was performed, accounting for the length of the sections in which the crop hedgerows were divided to extract the geometric parameters. Measurements from the UAV and the MTLS were highly correlated (R2 from 0.82 to 0.94) when considering the data from the three crops together, and the correlations were higher when analysing longer row sections. The canopy geometric parameters estimated using the MTLS-LiDAR always had higher values than those from the UAV-DAP. The results presented in this work provide useful data for a more informed selection of technological approaches for 3D crop characterization in Precision Fruticulture and high-throughput phenotyping., This research was funded by the projects DECIMAL (PID2020-113229RB-C41), PRODIGIA (PID2020-113229RB-C44), AGVANCE (AGL2013-48297-C2-2-R), and PAgPROTECT (PID2021-126648OB-I00), from the Spanish Ministry of Science and Innovation (MCIN/AEI/10.13039/501100011033). The authors also wish to thank Codorníu SA, celler Raimat and IRTA for having allowed the use of their vineyards, pear and peach orchards to conduct the trials described in this work.