With the availability of various satellite data sources such as Landsat and Sentinel images, use of spectral data and its application in agriculture are currently being tried and tested in the Philippines. These moderate to high resolution satellite imageries provide spectral variability that allow discrimination between healthy and infested plants which are now used in surveillance, early detection and monitoring of various pests and diseases of economically important crops in the Philippines. Another remote sensing technology that is recently tried under Philippine condition is the use of Light Detection and Ranging (LiDAR). This technology collects terrain surface information that can be used in classification and mapping. This technology was mainly used to determine the spatial distribution of high value crops in the agroecosystems, aquatic resources, hydrological resources, forest resources and renewable energy resources in the form of maps which were turned over to the Local Government Units (LGUs) and policy makers in the country to make them aware on the existence and inventory of different resources in their locality. On the other hand, the UAV or Unmanned Aerial Vehicles are now gaining popularity in the Philippines. High resolution multispectral imageries captured by this vehicle were used in providing updated geo-information and plant health status of the target areas by conducting image analysis and classification processes. The rotary type UAVs are now being tested in some research agencies in the country for rice seeding and spraying of pesticides. Big plantations run by multinational companies are using drones in spraying pesticides in lieu of airplanes. While the fixed wing drones are now used in surveillance, detection with high accuracy, monitoring, damage and, assessment of onion armyworm (Spodoptera exigua H.) infestation which is currently the most destructive insect pest of onion in the Philippines. Same is true on the anthracnose-twister which currently is the most destructive disease of onion in the Philippines. All the generated maps specially the maps that show where the pests or diseases were early detected were immediately turned over to the policy makers, Local Government Units (LGUs) and most especially to the local farmers. It is used as an early warning tool to forewarn the community of the impending pest or disease outbreak in the area.
Field spectroscopy is another promising tool that provides great advantage in crop production and protection. It offers a wide range of application in agriculture such that it can improve and replace the traditional method of obtaining the physiological and biochemical parameters of crops which is mainly based on taking physical samples from the fields and analysis in the laboratory but instead it can generate fast results in a non-destructive and less labor intensive way. A recent countrywide project utilized this tool and collected different spectral signatures to develop a spectral library database of the major crops in the country, the output of the project was intended to ease identification and characterization of crops using remote sensing data. Recent studies involving field spectral measurements were also conducted in analyzing and spectral characterization of Coconut Scale Insect (CSI) infestation in coconut, another is in Bacterial Leaf Blight (BLB) which is also a major disease of rice in the Philippines and these information were also applied in early detection, surveillance of other pests and diseases and even in damage assessment in communities, be they at the regional and national levels. From the agriculture point of view, the advantages of using remote sensing technology will far outweigh the investments by the government and the farmers as more and more stakeholders will adopt to the technology. Its application is not only confined to present subject matter, as a matter of fact there are those that are already planning to use it in energy potential determination of a particular crop, flood vulnerability assessment, crop suitability assessment, biodiversity assessment, soil health and even in forest health status determination and many more.
Keywords: Remote sensing, UAV, spectroscopy, agriculture, pests
The ever-increasing world population needs to meet the growing demand for agricultural products. To ensure global food supply, investigation and monitoring of agronomic parameters of crops is a necessary task (Mulla, 2013). Over the last decade, there has been an increasing interest among growers in developed countries in adopting precision agriculture, a crop-management based system that is based on spatial and temporal heterogeneity of soil and crops within the field (Stafford, 2000). Precision agriculture or site-specific management involves the integration of new technologies like Geographic Information Systems (GIS), Global Positioning Systems (GPS) and Remote Sensing (RS) technologies to allow farm producers to manage within field variability to maximize the cost benefit ratio, rather than using the traditional whole-field approach (Brisco, et.al. 1998). This approach recognizes the inherent spatial variability associated with crop growth and uses this information to prescribe the most appropriate management strategy on a site-specific basis. The driving force behind precision agriculture is the economic optimization of crop production. Intensive management of soil, water and pesticide inputs which has the potential to reduce costs by using inputs more efficiently. The site-specific application of these inputs reduces producer costs and minimizes the environmental impacts associated with chemical use because the chemical inputs can be applied at a correct rate only to the sites affected (Petersen, 1991).
Remote sensing is the science and technology of making inferences about material objects from measurements made at a distance without coming into physical contact with the objects under study (Menon, 2012.) The carrier of information in remote sensing is electromagnetic radiation, which travels in vacuum at the speed of light in the form of waves of different lengths. The most useful wavelengths in remote sensing cover visible light (VIS), and extend through the near (NIR) and shortwave (SWIR) infrared, to thermal infrared (TIR) and microwave bands (Wojtowicz, 2016). Remote sensing has a wide array of application in agriculture and is now slowly being adopted in developing countries like the Philippines. It is being used in agriculture resources monitoring, crop yield estimation based on vegetation indices, soil health monitoring, insect pests and diseases surveillance and detection, environmental disaster damage assessments and the like.
Satellite imageries that are used in remote sensing, (i.e. from moderate to high resolution) are becoming more available due to the increasing number of satellite missions dedicated to remote sensing. On the other hand, UAV or Unmanned Aerial Vehicle is another tool that can be used to generate high resolution images for specific location and can provide an access to immediate real-time quality data. Another reliable tool that goes hand in hand with remote sensing is the science of Spectroscopy that offers an alternative method of analysis, especially in agriculture because of its fast, reliable and non-destructive nature of analysis. Data generated out of using these tools can be used in a wide array of applications such as in agricultural production, resources management as well as in environmental monitoring.
This paper presents the current use and application of remote sensing in the Philippines particularly in the field of agriculture. It also discusses the challenges, trends and future prospects of the technology in the country.
Remote sensing trough LiDAR
Light Detection and Ranging (LiDAR) is a powerful remote sensing technology used in the acquisition of the terrain surface information, classification and extraction (Smadzadegan et al, 2010). On the other hand, the combination of high-resolution aerial imageries and LiDAR data provide the user an ease in separating different land cover classification of certain area to be mapped. Land cover classification is important in project planning and land use planning, and from time to time, land cover conversion into different land uses such as agricultural land converted into commercial establishment and just recently it is used in the inventory of agricultural crops planted in a particular area. The emergence of new spatial data acquisition systems such as Light Detection and Ranging (LIDAR) and Airborne Radar Interferometry (INSAR) presents complementary or alternative solutions to the acquisition of spatial information unanswered by existing technologies such as aerial photography or satellite imagery. The coverage and accuracy of topographic data extracted by these systems, complemented by the features detected by an onboard digital aerial camera, provide rich information that would greatly benefit agencies using spatial data.
A recent nationwide project initiated by the Department of Science and Technology (DOST) in the Philippines known as Phil-Lidar 2 whereby LiDAR technology was used by different regional universities throughout the country to determine the vulnerability of the agriculture sector towards climate change and to determine the spatial distribution of high value crops in the agroecosystems, aquatic resources, hydrological resources, forest resources and renewable energy resources. LiDAR data and high-resolution images were used to derive different derivatives to produce high resolution maps. Derivatives like DTM, DSM, nDSM, intensity, number of returns, GRVI and HSV were generated using LasTools, ArcGIS and ENVI softwares. These derivatives were imported in eCognition software where the main analysis was carried out using Object Based Image Analysis (OBIA). Land cover were delineated and classified into pre-defined classes to achieve proper detection features for land cover classification mapping. ArcGIS was used for the selection of validation sets. Mapping using OBIA was proven to be an effective tool in deriving land cover classification information for agriculture and aquaculture resources based on the available LIDAR data and aerial images. The produced maps have an accuracy ranging from 85% to 90% and an average Kappa coefficient of 0.87 (Figure 1).
The technology was also used to assess the vulnerability of agricultural resources to floods and droughts, the vulnerability of coastal resources to climate change hazards as well as in determining the most suitable site for planting a particular crop in the area (Figure 2).
UAV in agriculture
Once best known as a military surveillance tool, the UAV has more recently become popular as a recreational devise and has also been put to good use including agriculture. The UAV have the potential to reduce time-consuming, cost –prohibitive and labor-intensive ground-based surveillance which includes monitoring hazardous environments, inspecting bridges and pipelines, deployment in search and rescue operations, making rapid damage assessment and even in human disease epidemiological studies. The use of unmanned aerial vehicles (UAVs), also known as drones, and connected analytics has a great potential to support and address some of the most pressing problems faced by agriculture in terms of access to actionable real-time quality data. Goldman Sachs predicts that the agriculture sector will be the second largest user of drones in the world in the next five years (Sylvester, 2018).
For the past decades, UAV was proven to be applicable in many technological spheres. The growing number of the world’s population and the fast-industrial development tend to overexpose the soils and arable fields. In order to meet the yield’s expectations nowadays, agriculture could even be a threat to the environment. That is why scientists must seek better solutions and reliable techniques to preserve the environment to increase the potential of agriculture in a sustainable way; one way to attain this goal is to continuously monitor what was happening in the field. One of the methods proposed to monitor vegetation is thru multispectral and thermal imagery in combination with UAV (Raeva et al., 2019).
The images captured by the satellite and UAVs are datasets that can be utilized in resource mapping. These datasets can generate Vegetation Indexes (VIs) which can be used to determine the health and strength of vegetation, vegetation density, with the aim to obtain those formulas that get more reliable information about vegetation based on remotely sensed values (Suarez et al. 2017).
Alberto et al., (2019b) used Unmanned Aerial Vehicle (UAV) to accurately identify and map the anthracnose-twister disease in onion (Figure 3). The manifestations of this disease in onion areas are very visible in aerial imageries captured by UAV’s, thus, these imageries were utilized in extracting infected onion areas in the fields. To map out the affected areas, Object Based Image Analysis (OBIA) was carried out in aerial imageries captured by the UAV’s. Vegetation indices generated from the RGB and NIR bands were used as image layers and the Support Vector Machine as the classifier. The Support Vector Machine (SVM) was used to generate geo-phytopathological maps showing the actual picture and health status of onion fields with 85+% accuracy. The OBIA using SVM was effective in extracting infected onion areas using different vegetation indices, thereby, creating geo-phytopathological maps pin pointing the infected and the non-infected fields in the areas. These, maps were turned over to the decision makers and extension workers to raise the level of awareness on the infestation and used as monitoring tool in disease spread prevention as well as in planning for disease and pesticide management and environmental protection.
Another study of Alberto et al., (2019c) used Unmanned Aerial Vehicle (UAV) in monitoring and mapping of armyworm infested onion areas (Figure 4). UAV captured imageries were processed using Pix4D-Mapper. Also, the index (Normalized Difference Vegetation Index (NDVI)) and orthomosaics of captured images were produced and analyzed onion fields. Similar approach was used in detection, monitoring and mapping of leaf miner infested onion areas in the Province of Nueva Ecija.
Remote sensing in insect pests and diseases surveillance
Satellite images provide spectral properties at the visible, near infrared and shortwave infrared regions which help in insect pest and disease detection due to the variability in the spectral reflectance depending on the condition of the plant. Recent study of Alberto et. al, (2018) evaluated the vegetation indices from Sentinel 2 imagery in detecting twister disease of onion. To provide timely and accurate early detection of twister disease of onion in the field, remote sensing was tried and tested using Sentinel 2 imageries. Vegetation indices (VIs) derived from the VIS-NIR region of the image were evaluated for their capability to detect twister disease. VIs were subjected to regression analysis to evaluate the relationship between vegetation indices and severity index of onion twister disease. Vegetation indices with strong relationship to twister disease were selected and further used in unsupervised ISODATA classification. Overall accuracy of classification generated from vegetation indices was calculated based on confusion matrix using ground truth points collected from field work to identify the most suitable index based on highest overall accuracy. It shows that NDVI and GNDVI has the highest coefficient of determination (R2) indicating a strong relationship with the disease severity. Results of the classification showed that GNDVI PSSRa and NDVI obtained the highest overall accuracy. This indicates that the 3 VIs can be used for detection of twister as it gives better discrimination and high accuracies. Hence VI’s generated from Sentinel 2 imagery has a potential to use in the management of twister disease of onion.
Another study of Alberto et al, (2019a) used remote sensing in early detection, spatial tracking and mapping of armyworm (Spodoptera exigua H.) in onion (Figure 5). The study covers the Province of Nueva Ecija which is the major onion producing province in the country with 18 of which are onion growing Municipalities. The products or maps generated shows the occurrence of armyworm in a particular location, incidence and damages of the infestation which is updated weekly, hot spot areas showing the concentration or intensity of the insect pest and as well as yield assessment maps were turned over to partner agencies and institution such as Municipal Agriculture Office (MAO), Provincial Agriculture Office (PAO) and Regional Crop Protection Center (RCPC) to help them in the decision and policy making, planning, identification of priority areas and managing the pest outbreak.
Plant disease prediction and risk maps
Maps are powerful tools to convey information to the farmers. Maps can be used for identification and exploration of dispersal pattern that can give a complete picture for monitoring and management of pests and diseases (Kishojini et al., 2018). Plant disease risk maps are particularly appropriate because of the high levels of spatial and temporal variations (Burdon and Thrall, 1999). Observations of historical or current disease severity could be one of the most useful predictors of future disease risk, particularly when the pathogen and the host were not limiting during previous evaluations. Maps of previously known locations for plant disease can be useful in decision-making on the introduction of new germplasm.
Alberto et al., (2018) generated disease risk map of anthracnose-twister of onion based on previous disease locations as a future predictor or as an early warning tool. The data of previous disease locations were utilized to generate prediction and disease risk maps through interpolation using Kriging model. Based on the results, the prediction map suggest that the anthracnose-twister disease of onion has the possibility to developed into an epidemic scale and the disease outbreak will most likely to occur in the southern part of Bongabon (Philippines). It shows that the southeastern part of Bongabon has a very high risk due to the high incidence rate (50.01% to 75.00%) on this area during the previous cropping seasons. Through these maps, preventive measures can be recommended and relayed at the earliest possible time to the farmers even though the disease have not yet occurred in the area (Figure 6). To mitigate the situation in these areas it is recommended to avoid using white onion varieties which is very susceptible to anthracnose-twister, and spray protectant fungicides 1 week after transplanting.
Ground based remote sensing/spectroscopy
A spectrometer (spectrophotometer, spectrograph or spectroscope) is an instrument used to measure properties of light over a specific portion of the electromagnetic spectrum, typically used in spectroscopic analysis to identify materials. The variable measured is mostly the light’s intensity but could also for instance be used in the polarization state (Hashim et. al., 2010). On the other hand, field spectroscopy is a technique used to measure the reflectance properties of vegetation, soils, rocks and water bodies in the natural environment, generally under solar illumination (Milton, 2003). Application of field spectroscopy in agricultural crops is important because it can provide data on plant characteristics such as leaf structure, nutrient and water status, leaf age, and leaf pigment. Field spectroscopy has certain advantage as compared to the traditional method of obtaining the physiological and biochemical parameters of crops which is mainly based on taking physical samples from the fields, and then measuring them using chemical methods in the lab which is time consuming, labor intensive, and destructive (Roth et al., 1989). In this context, spectroscopy, a non-destructive association with rapid measurements of green biomass can be done in order to produce an index for defining the vegetation (Romano et al, 2011) whereas, different groups of pigments absorb special light wavelengths.
Spectral characterization of bacterial leaf blight (BLB) of rice through spectroscopy and remotely sensed multi-spectral imagery was conducted by Alberto et al. (2017). In this study, spectroscopy and multi-spectral imaging were applied to rapidly and non-destructively characterized BLB-infected rice areas in the field. Three sample leaves were collected according to the degree of disease severity, from normal (healthy), slight, moderate to severe, as described by Zhao et al. (2013) with a total number of 12 samples. The spectra collected were in visible to near infrared (NIR) range. The data acquired were processed using spectral calculator to obtain the ratio between reflectance and wavelength within the electromagnetic spectrum. Multi-spectral Landsat 8 Operational Land Imager (OLI) imagery (level 1) was downloaded in Geotiff format and pre-processed in ENVI software using radiometric calibration. Leaf spectroscopy analysis indicated that there is a greater variation between healthy and BLB infected rice plant whereby the most responsive wavelength was observed between 650 to 680 nanometer (red region) and 749 to 769 (Near-Infrared region). Wavelengths from 630 - 670 nm (red region) were found to be statistically separable in Landsat 8 image as an indicator for BLB infection. The study suggests that both leaf spectroscopy and multi-spectral images have the capability to discriminate healthy and BLB-infected rice from aerial imagery (Figure 7). These findings show that there is a big possibility of using multi-spectral images in estimating BLB infection in a larger scale. This can also be used to assess the proper timing of applying interventions or in estimating potential losses due to BLB infection, thereby allowing precision disease management. The study also suggests to use high resolution images in further characterizing different severity levels of BLB infection and to avoid spectral mixing caused by medium resolution multispectral image like Landsat 8 OLI.
Paringit et al., (2014) also conducted spectral characterization of coconut scale insect (CSI) from field spectroradiometric measurements and high-resolution superspectral imageries. They examined the capability of optical remote sensing spectral techniques to detect the presence of CSI from field spectral measurements high resolution multispectral satellite imagery (HRMI) of coconut planted areas at least for the pilot sites examined. They did this because the infestation caused mortality of coconut trees that reduced the productivity and threatened the coconut industry in the country. Remote sensing techniques are explored as a means to rapidly survey and monitor the CSI problem. Field spectral measurements were conducted to analyze the spectral features of coconut leaves within different levels of infestation (low, moderate and severe), tree trunk and stand understory (e.g. grass). Worldview-2 images of coconut stands taken from two different dates more than a year apart covering the affected areas were calibrated, co-registered and analyzed. Spectral signatures of coconut of various degrees of infestation are fairly distinct and distinguishable in the 8-band Worldview-2 satellite imagery particularly pronounced in the NIR-1 and NIR-2 bands followed by Red Edge band. Ground truth data is consistent with the findings of satellite image analysis. The changes in these particular bands are also pronounced in the December 2012 versus January 2014 imagery. Spectral characteristics of affected coconut plants can be used as indicators to rapidly detect distribution, level and extent of infestation.
Future prospects of remote sensing in the Philippines
To realize the potential that remotely sensed data offers, there were already softwares that can be used to extract, analyze patterns that are not clearly discernable to the human eye. This offers a window into the growth cycle that enables early stress and disease detection and provides information so interventions can be implemented before a crop becomes permanently impaired. Regardless of the technological advances, however, the bottom line here is how the farmers can utilize these technologies to make better crop management decisions that would translate to increase returns.
Although the use of remote sensing in agriculture in the Philippines is still in its infancy stage, and considering present trends in the development of advanced UAV and sensor systems, there is no doubt that more agricultural producers and researchers will utilize these technologies in agricultural production in the future. The increased potential for incorporating innovative technologies in agriculture is already felt in the Philippines by the presence of startup companies that are engaged in developing farm management softwares, drones and sensor technologies and predictive data analytics. Coupled with these scenarios is the availability of imagery providers which provide temperature, vegetation indices and true color imagery which are vital in agriculture. It is now gaining popularity as various institutions are now looking into this technology to be used in different agricultural researches and production initiatives. However, there are problems and limitations that still needs to be addressed in order for this technology to establish its foothold and adopt this technology as mainstream tool in agriculture such as: variability in the reflectance caused by solar illumination, low and moderate resolution images from satellite image providers, very expensive softwares and hardwares and limited number of people trained this area. Another problem which needs to be improved is the reliability or accuracy of the detection algorithm since the freely available satellite images are low to moderate in resolutions where spectral missing is common and sometimes introduce false signals or “noise”. Also, processing massive data and delivering this information in a timely fashion and in a way that the grower can minimize crop inputs and maximize yields.
Drones for pesticide spraying and seeders are now being introduced by various private companies to farmers which offer a more convenient way of farming. Lastly, the launching of the microsatellite of the Philippines which is known as Diwata 2 which offers a high-resolution images with 5 meters resolution thru its High Precision Telescope (SMI) camera.
Though there is no perfect technology, the positive outlook of using these technologies will far outweigh the investments made by the government as the data that will be generated will assist in the development of different agricultural maps which will offset farm expenses. This map-based approach will enable growers to precisely locate-pests and disease prone areas, nutrient deficient soil areas, flood prone areas and crop suitability areas. Developing maps in conjunction with yield, nutrient, disease, weed and other pest maps will give a comprehensive picture of the entire field that a grower and decision makers can rely upon for executing management decisions that are timely, eco-friendly and economical. Moreover, it will enhance crop monitoring and management which would allow precise application of inputs thereby increasing economic returns and crop yields (Mattupalli,et.al.,2017).
Alberto, R.T., M.F Isip. and A.R. Biagtan, 2019a. Hot spot area analysis of onion armyworm outbreak in Nueva Ecija using geographic information system. Spatial Information Research. ISSN 2366-3286. DOI 10.1007/s41324-019-00266-0
Alberto, R.T., M.F. Isip, and A.R. Biagtan. 2019b. Extraction of onion fields infected by anthracnose-twister disease in selected municipalities of Nueva Ecija using object-based image analysis. 51st Anniversary and Annual Scientific Conference of the Pest Management Council of the Philippines. Coron, Palawan, Philippines
Alberto, R.T., R. C. Tagaca, K. C. Manipon and K. T. F. Alejandro. 2019c. Use of Unmanned Aerial Vehicle (UAV) and Pix4D-Mapper in Monitoring and Mapping of Armyworm Infested Onion Areas in Brgy. Lusok, Bongabon, Nueva Ecija. 51st Anniversary and Annual Scientific Conference of the Pest Management Council of the Philippines, Coron, Palawan, Philippines.
Alberto, R.T., M.F. Isip, and A.R. Biagtan. 2018. Disease risk map of anthracnose-twister of onion based on previous disease locations as a future predictor. Spat. Inf. Res. https://doi.org/10.1007/s41324-018-0229-4
Alberto, R. T., M.F. Isip, D.C.J. Pangilinan. 2017. Spectral characterization of bacterial leaf blight (BLB) of rice through spectroscopy and remotely sensed multi-spectral imagery. 49th Anniversary and Annual Scientific Conference of the Pest Management Council of the Philippines. Crown Regency Resort and Convention Center, Boracay Island, Aklan, Philippines.
Brisco, B., R.J., T. Brown, T. Hirose, T. McNairn, and H. Staenz, Precision agriculture and the role of remote sensing: A review. Can. J. Remote Sens. 24: 315–327
Burdon, J. J., & Thrall, P. H. 1999. Spatial and temporal patterns in co-evolving plant and pathogen associations. American Naturalis. 153: S15.
Hashim,H., M.A. Haron, F.N. Osman, and S.A.M. Junid. 2010. Classification of rubber tree leaf disease using spectrometer. Fourth Asia International Conference on Mathematical/Analytical Modelling and Computer Simulation.
Kishojini, P., K., Pakeerathan, and G. Mikunthan. 2018. GPS based density and distribution mapping and composting a sustainable approach for monitoring and managing parthenium (Parthenium hysterophorus L.) in northern Sri Lanka. International Journal of Agriculture and Forestry, 8(4): 160–170. https://.org/10.5923/j.ijaf.20180804.05.
Mattupalli, C., M.R. and C.A. Young. 2017. Integrating geospatial technologies and unmanned aircraft systems into the grower’s disease management toolbox. APS Features.doi: 10.1094/APSFeature-2017-7.
Marek, W, A.Wójtowicz and J. Piekarczyk. 2016. Application of remote sensing methods in agriculture. Communications in Biometry and Crop Science. 11(1): 31–50.
Menon, A. R. R. 2012. Remote Sensing Application in Agriculture and Forestry. Proceedings of the Kerala Environment Congress.
Milton, E. J. 2003. Field spectroscopy. Geoinformatic – Vol 1 –Field Spectroscopy
Mulla, D.J. 2013. Twenty five years of remote sensing in precision agriculture: Key advances and remaining knowledge gaps. Biosystems Engineering.114 (4):358-371.
Paringit, E., J. Fabila, J. Ilagan, M. J., Cruz, C., and S. Samalburo. 2014. Spectral characterization of coconut scale insect (CSI) from field spectroradiometric measurements and high-resolution superspectral imagery. 35th Asian Conference on Remote Sensing. Manila, Philippines.
Raeva, P.L., J. Šedina, and D. Lesk, A., 2019. Monitoring of crop fields using multispectral and thermal imagery from UAV. European Journal of Remote Sensing, 52, 192–201
Romano, G., S., W. Zia, W. Spreer, C. Sanchez, J. Cairns, J.L. Araus and J. Muller. 2011. Use of thermography for high throughput phenotyping of tropical maize adaptation in water stress. Computers and Electronics in Agriculture, 79:67–74.
Roth, G.W., R.H. Fox, and H.G. Marshall. 1989. Plant tissue tests for predicting nitrogen fertilizer requirements of winter wheat. Agronomy Journal. 81: 502-507.
Smadzadegan F., B. Bigdeli and P. Ramzi. 2010, Classification of LiDar data based on Multi-class SVM, Commission VI, WG VI/4, Tehran, Iran
Suarez, P.S., A.B. Vintimilla. 2017. Learning image vegetation index through a conditional generative adversarial network. 10.1109/ETCM.2017.8247538.
Sylvester G. 2018. E-agriculture in action: Drones for agriculture. Food and Agriculture Organization of the United Nations and International Telecommunication Union. ISBN 978-92-5-130246-0s
Date submitted: October 29, 2019