Title: Usage of Machine Learning Algorithms on Precision Agriculture Applications

Abstract: Agricultural monitoring and analysis of data to be used in management decisions to increase the quality, profitability, sufficiency, continuity and efficiency of agricultural production is called Precision Agriculture.[1]Precision Agriculture technologies aim to help the farmers with the decision making process by providing them information and control over their land, crop status and environment using remote sensing systems. Remote sensing systems use multispectral cameras to gather information, which filter different wavelengths of light in separate bands. Vegetation indices derived from the spectral bands of the remote sensing systems carry useful information about crop characteristics such as nitrogen content, chlorophyll content and water stress which supports the farmers to plan irrigation and pesticide spraying processes without the need of manual examination, providing a cost and time-efficient solution. This study aims to explore three specific Precision Agriculture applications, such as crop segmentation, illness detection and yield prediction on olive trees in Manisa, Turkey by using machine learning algorithms. Using the spectral band information gathered from an Orange-Cyan-NIR (OCN) camera embedded UAV system, vegetation health index was calculated and the data was preprocessed by segmentating the tree pixels from background based on those values using MiniBatchKMeans algorithm. Optimal features were selected based on accuracy comparison for yield and disease predictions. A Decision Tree Regressor (DTR) model was trained for yield prediction while a Random Forest Classifier (RFC) model was trained for disease prediction. The results showed that crop segmentation had an accuracy rate of 0.85-0.95, while DTR and RFC models had an R2 score of 0.99 and accuracy rate of 0.98 respectively, which displayed the importance and usefulness of vegetation indices.

Keywords: Machine learning; artificial intelligence; precision agriculture; yield prediction; disease prediction; crop segmentation

Başlık: Hassas Tarım Uygulamaları Üzerinde Makine Öğrenmesi Teknikleri Kullanımı

Özet: Hassas Tarım, tarımsal üretimin devamlılığı, kalitesi, yeterliliği, karlılığı ve verimliliğini arttırmak adına verilen yönetimsel kararlarda tarımsal gözetim ve veri analizi teknikleri kullanılmasıdır. Bu teknolojiler, çiftçilere ürünleri, toprakları ve çevreleri hakkında uzaktan algılama sistemleri kullanıp kontrol ve bilgi olanağı sağlayarak karar verme süreçlerinde yardım etmeyi hedefler. Uzaktan algılama sistemleri bilgi toplamak için ışığın farklı dalga boylarını ayrı bantlarda filtreleyen multispektral kameralar kullanır. Spektral bantlardan elde edilen bitki örtüsü indeksleri, azot, klorofil ve su stresi gibi ürün özellikleri hakkında bilgi taşır ve bu bilgiler çiftçilerin, arazide herhangi bir ölçüme gerek kalmaksızın, sulama ve ilaçlama kararlarında yardımcı olacak niteliktedir. Bu çalışmada hassas tarım uygulamalarında makine öğrenmesi tekniklerinin kullanımının keşfi hedeflenmiş olup, Türkiye’nin Manisa ilçesindeki zeytin ağaçlarına yoğunlaşılmıştır. Bir İHA sistemine entegre edilmiş OCN kamerasının spektral verilerini kullanarak bitki örtüsü sağlık endeksi hesaplanmış ve veride bu endeksi baz alarak ağaç pikselleri diğerlerinden MinBatchKMeans algoritmasıyla ayrıştırılmıştır. Rekolte ve hastalık tahmini için en uygun öznitelikler doğruluk oranı karşılaştırmasına göre seçilmiştir. Rekolte verisi için Karar Ağacı Regresyon (KAR) modeli eğitilmiş, hastalık tahmini için Rastgele Orman Sınıflandırma (ROS) modeli eğitilmiştir. Ağaç piksel ayrıştırma sonucunun doğruluk oranı 0.85 ve 0.95 arasında belirlenmiş, KAR algoritmasının R2 puanı 0.99 ve ROS algoritmasının doğruluk oranı 0.98 olarak hesaplanmıştır, bu da bitki örtüsü indekslerinin önemi ve kullanışlılığını vurgulamaktadır.

Anahtar kelimeler: Makine öğrenmesi; yapay zeka; hassas tarım; rekolte tahmini; hastalık tahmini; bitki ayrıştırma

• International Society of Precision Agriculture. Workshop Digital Agriculture Profile for Turkey. ISPA Newsletter 2019, Retrieved from https://www.ispag.org/about/newsletters?preview=90
• Bai XD, Cao ZG, Wang Y, Yu ZH, Zhang XF, Li CN. Crop segmentation from images by morphology modeling in the CIE L*a*b* color space. Computers and Electronics in Agriculture 2013; 99: 21–34.
• Torres-Sanchez J, Lopez-Granados F, Pena JM. An automatic object-based method for optimal thresholding in UAV images: Application for vegetation detection in herbaceous crops. Computers and Electronics in Agriculture 2015; 114: 43–52.
• Li Y, Huang Z, Cao Z, Lu H, Wang H, Zhang S. Performance evaluation of crop segmentation algorithms. IEEE Access 2020; 8: 36210–36225.
• Lottes P, Khanna R, Pfeifer J, Siegwart R, Stachniss C. UAV-based crop and weed classification for smart farming. In 2017 IEEE International Conference on Robotics and Automation (ICRA), May 29-June 3, 2017, Singapore.
• Pantazi XE, Moshou D, Alexandridis T, Whetton RL, Mouazen AM. Wheat yield prediction using machine learning and advanced sensing techniques. Computers and Electronics in Agriculture 2016; 121: 57–65.
• Ashapure A, Oh S, Marconi TG, Chang A, Jung J, Landivar J, Enciso J. Unmanned aerial system based tomato yield estimation using machine learning. In SPIE Proceedings Vol. 1100800: Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping IV, 2019, Baltimore, Maryland, USA, pp. 171-180.
• Panda SS, Ames DP, Panigrahi S. Application of vegetation indices for agricultural crop yield prediction using neural network techniques. Remote Sensing 2010; 2(3): 673–696.
• You J, Li X, Low M, Lobell D, Ermon S. Deep Gaussian process for crop yield prediction based on remote sensing data. In Proceedings of the THirty-First AAAI Conference on Artificial Inteligence, 2017, pp. 4559-4566.
• Ferentinos KP. Deep learning models for plant disease detection and diagnosis. Computers and Electronics in Agriculture 2018; 145: 311–318.
• Khamparia A, Saini G, Gupta D, Khanna A, Tiwari S, De Albuquerque VHC. Seasonal crops disease prediction and classification using deep convolutional encoder network. Circuits, Systems, and Signal Processing 2020; 39: 818-836.
• Pantazi XE, Tamouridou AA, Alexandridis TK, Lagopodi AL, Kontouris G, Moshou D. Detection of Silybum marianum infection with Microbotryum silybum using VNIR field spectroscopy. Computers and Electronics in Agriculture 2017; 137: 130–137.
• Chung CL, Huang KJ, Chen SY, Lai MH, Chen YC, Kuo YF. Detecting Bakanae disease in rice seedlings by machine vision. Computers and Electronics in Agriculture 2016; 121: 404–411.
• MAPIR. OCN filter improves results compared to RGN filter. Retrieved from https://www.mapir.camera/pages/ocn-filter-improves-contrast-compared-to-rgn-filter
• Perko R, Raggam H, Gutjahr K, Schardt M. The capabilities of TerraSAR-X imagery for retrieval of forest parameters. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives, July 5-7, 2010, Vienna, Austria, pp. 452-456.
• Rouse JW, Haas RH, Schell JA, Deering DW. Monitoring vegetation systems in the great plains with ERTS. In Proceeding of Third Earth Resources Technology Satellite Symposium, USA, 1974, vol. 1, pp. 309-317.
• Huete AR. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment 1988; 25(3): 295–309.
• Liu HQ, Huete A. A feedback based modification of the NDVI to minimize canopy background and atmospheric noise. IEEE Transactions on Geoscience and Remote Sensing 1995; 33(2): 457–465.
• Penuelas J, Gamon JA, Griffin KL, Field CB. Assessing community type, plant biomass, pigment composition, and photosynthetic efficiency of aquatic vegetation from spectral reflectance. Remote Sensing of Environment 1993; 46(2): 110–118.
• Zhao B, Duan A, Ata-Ul-Karim ST, Liu Z, Chen Z, Gong Z, Zhang J, Xiao J, Liu Z, Win A, Ning D. Exploring new spectral bands and vegetation indices for estimating nitrogen nutrition index of summer maize. European Journal of Agronomy 2018; 93: 113–125.
• McFeeters SK. The use of the normalized difference water index (NDWI) in the delineation of open water features. International Journal of Remote Sensing 1996; 17(7): 1425–1432.
• Wilson EH, Sader SA. Detection of forest harvest type using multiple dates of Landsat TM imagery. Remote Sensing of Environment 2002; 80(3): 385-396.
• Aydemir M. Agricultural pest control technical instructions. Republic of Turkey Ministry of Agriculture and Forestry, General Directorate of Agricultural Research and Policies 2008; 5: 140-141.