Characterization of peach tree crown by using high-resolution images from an unmanned aerial vehicle

In orchards, measuring crown characteristics is essential for monitoring the dynamics of tree growth and optimizing farm management. However, it lacks a rapid and reliable method of extracting the features of trees with an irregular crown shape such as trained peach trees. Here, we propose an efficient method of segmenting the individual trees and measuring the crown width and crown projection area (CPA) of peach trees with time-series information, based on gathered images. The images of peach trees were collected by unmanned aerial vehicles in an orchard in Okayama, Japan, and then the digital surface model was generated by using a Structure from Motion (SfM) and Multi-View Stereo (MVS) based software. After individual trees were identified through the use of an adaptive threshold and marker-controlled watershed segmentation in the digital surface model, the crown widths and CPA were calculated, and the accuracy was evaluated against manual delineation and field measurement, respectively. Taking manual delineation of 12 trees as reference, the root-mean-square errors of the proposed method were 0.08 m (R^(2) = 0.99) and 0.15 m (R^(2) = 0.93) for the two orthogonal crown widths, and 3.87 m2 for CPA (R^(2) = 0.89), while those taking field measurement of 44 trees as reference were 0.47 m (R^(2) = 0.91), 0.51 m (R^(2) = 0.74), and 4.96 m2 (R^(2) = 0.88). The change of growth rate of CPA showed that the peach trees grew faster from May to July than from July to September, with a wide variation in relative growth rates among trees. Not only can this method save labour by replacing field measurement, but also it can allow farmers to monitor the growth of orchard trees dynamically.

Easy domain adaptation method for filling the species gap in deep learning-based fruit detection

Fruit detection and counting are essential tasks for horticulture research.With computer vision technology development,fruit detection techniques based on deep learning have been widely used in modern orchards.However,most deep learning-based fruit detection models are generated based on fully supervised approaches,which means a model trained with one domain species may not be transferred to another.There is always a need to recreate and label the relevant training dataset,but such a procedure is time-consuming and labor-intensive.This paper proposed a domain adaptation method that can transfer an existing model trained from one domain to a new domain without extra manual labeling.The method includes three main steps:transform the source fruit image(with labeled information)into the target fruit image(without labeled information)through the CycleGAN network;Automatically label the target fruit image by a pseudo-label process;Improve the labeling accuracy by a pseudo-label self-learning approach.Use a labeled orange image dataset as the source domain,unlabeled apple and tomato image dataset as the target domain,the performance of the proposed method from the perspective of fruit detection has been evaluated.Without manual labeling for target domain image,the mean average precision reached 87.5%for apple detection and 76.9%for tomato detection,which shows that the proposed method can potentially fill the species gap in deep learning-based fruit detection.

Responses of Root Hydraulic Properties and Transpirational Factors to a Top Soil Drying in <i>Cajanus cajan</i>and <i>Sesbania sesban</i>

Responses of leaf area (LA), stomatal conductance (gs), root length (RL) and root hydraulic conductance per unit of root length (Lpunit) to top soil dryness were investigated. Pigeon pea (Cajanus cajan) and sesbania (Sesbania sesban) were grown in a vertical split-root system. From sixty-six days after sowing, the top soil was dried while the bottom soil was kept wet. Pigeon pea increased LA while maintaining leaf water potential (ΨL) by reducing gs. Increased transpirational demand through canopy development was compensated for by increasing water extraction in the bottom soil. This was achieved by increasing not only RL but also Lpunit. Sesbania kept constant levels of gs, causing a transient reduction of ΨL. ΨL of sesbania was, then, recovered by increasing only RL, but not Lpunit, in the bottom soil while suspending LA extension, suggesting that sesbania regulated only the root area to LA ratio. This study demonstrated a species-specific significance of Lpunit and coordination among Lpunit, RL, gs and LA in exploitation of wet-deeper soils in response to top soil dryness.

Evaluation of Soil Water Management Difference in Mango Orchards between Thailand and Japan

The objective of this study is to evaluate the difference of the soil water management in mango orchards between the varieties of “Irwin” in Japanand “Nam Dok Mai” inThailand. Field observations were conducted in mango orchards in Okinawa, Japan and Phrao, Thailand to clarify the water management practices. Measurement of the hourly soil water content in Phrao indicated that the irrigation was scarce and the volumetric water content in the soil was maintained almost constant. in the flowering season. This can be the farmers’ practice for flower induction. After the flowering season, irrigation was frequent in order to produce the large fruit. In the harvest season, the soil water content was relatively high because of frequent irrigation and rainfall. In Okinawa, the volumetric water content was maintained at the same level in a relatively deep layer. The result at the5 cmdepth indicated that the farmer carefully controlled the soil water content. In the flowering season, the soil water content was relatively low. While the orchard was managed empirically, the volumetric water content near the soil surface was maintained over 25% during the harvest season. This result indicates that the farmer performed the good soil water management to enhance mango fruit quality even without technical measurement. A numerical model describing the soil water and heat transfers was introduced to predict the farmer’s empirical soil water management in Okinawa. Using the meteorological data in March 2010, the irrigation regime was predicted using the simulated soil water content. In the flowering season, the farmer irrigated when the soil surface water content reached 14%. Based on this criterion for the empirical soil water management, the simulation result indicated that the farmer irrigated four times in this period. The numerical model presented here can be useful for evaluating the differences in water management practices of local farmers.

Cultivation of <i>Erianthus</i>and Napier Grass at an Abandoned Mine in Lampung, Indonesia

The production of cellulosic bioethanol from non-edible plants is drawing increasing attention, as it potentially avoids food-fuel competition. Because growing such plants on farmland indirectly reduces food availability, the plants should be grown on marginal, non-arable lands. In this study, we evaluated the growth of cellulosic energy crops at a former mining site in Indonesia. This mine was abandoned because it contained few mineral deposits, and exposed subsoils rather than toxic soils prevented revegetation. In the first trial, growths of two energy plant species Erianthus spp. and Napier grass (Pennisetum purpureum) were compared with that of maize (Zea mays) at the mine site and a nearby degraded farm. Erianthus and Napier grass produced 11.7 and 22.5 t·ha-1 of shoot dry matter at 8 months after planting (MAP) in the farm respectively while maize plants failed to establish, but none of the three species grew at the mine. In the second trial, two-week-old seedlings of Erianthus and Napier grass rather than stem cuttings as used in the first trial were planted at the mine site. Erianthus and Napier grass produced 16.3 and 24.0 t·ha-1 of shoot dry matter over the course of 18 months, respectively. Application of organic fertilizer significantly increased shoot dry matter to 18.9 and 39.6 t·ha-1 in Erianthus and Napier grass, respectively. During the 18-month growth period, both of the energy plants significantly increased soil carbon at the 0 - 0.3 m depth from 0.33% to 1.15% - 1.23% when chemical fertilizer was applied and to 0.67% - 0.69% when both chemical and organic fertilizers were applied. From 0 - 5 MAP, soil surface level dropped by 28.0 - 34.7 mm in plots without plants due to soil erosion. In contrast, both of the energy plants significantly reduced the drop of soil surface level to 16.0 - 19.3 mm in plots with chemical fertilizer alone and to 18.0 - 20.7 mm in plots with chemical and organic fertilizers. Proportions of small soil particles, that would be easily detached and transported by water flow compared with large particles, were larger in the planted plots than the no-plant plots at 16 MAP. The results suggest that successful cultivation of energy plants on abandoned mine sites is possible, particularly if seedlings are transplanted and the crops are fertilized with organic fertilizer. In addition, the cultivation of Erianthus and Napier grass has positive impacts on soil quality that may contribute to their sustainability as crops and to the conservation of the local ecosystem.

Root Morphology and Anatomy of Field-Grown Erianthus arundinaceus

Erianthus species are perennial C4 grasses with such high biomass productivity and high tolerance to environmental stresses that they can be grown in marginal land to supply raw material for cellulosic bioethanol. Because high biomass production and strong tolerance to environmental stresses might be based on their large and deep-root system, we closely examined the morphology and anatomy of roots in first-year seedlings of field-grown Erianthus arundinaceus. The deep-root system of E. arundinaceus consists of many nodal roots growing with steep growth angles. Diameter of nodal roots with large variations (0.5 - 5 mm) correlates with the size and number of large xylem vessels. The microscopic observation shows that the nodal roots with dense root hairs developed soil sheath, hypodermis with lignified sclerenchyma in the outer cortex, and aerenchyma in the mid-cortex. In addition, starch grains were densely accumulated in the stele of nodal roots in winter. In the first year, E. arundinaceus developed less lateral roots than other reported grass species. The lateral roots formed a large xylem vessel in the center of the stele and no hypodermis in the outer cortex. Morphology and anatomy of E. arundinaceus root were discussed with reference to strong tolerance to environmental stresses.

A Weakly Supervised Deep Learning Framework for Sorghum Head Detection and Counting

The yield of cereal crops such as sorghum(Sorghum bicolor L.Moench)depends on the distribution of crop-heads in varying branching arrangements.Therefore,counting the head number per unit area is critical for plant breeders to correlate with the genotypic variation in a specific breeding field.However,measuring such phenotypic traitsmanually is an extremely labor-intensive process and suffers from low efficiency and human errors.Moreover,the process is almost infeasible for large-scale breeding plantations or experiments.Machine learning-based approaches like deep convolutional neural network(CNN)based object detectors are promising tools for efficient object detection and counting.However,a significant limitation of such deep learningbased approaches is that they typically require a massive amount of hand-labeled images for training,which is still a tedious process.Here,we propose an active learning inspired weakly supervised deep learning framework for sorghum head detection and counting from UAV-based images.We demonstrate that it is possible to significantly reduce human labeling effort without compromising final model performance(𝑅2 between human count and machine count is 0.88)by using a semitrained CNN model(i.e.,trained with limited labeled data)to perform synthetic annotation.In addition,we also visualize key features that the network learns.This improves trustworthiness by enabling users to better understand and trust the decisions that the trained deep learning model makes.

Robust Surface Reconstruction of Plant Leaves from 3D Point Clouds

The automation of plant phenotyping using 3D imaging techniques is indispensable.However,conventional methods for reconstructing the leaf surface from 3D point clouds have a trade-off between the accuracy of leaf surface reconstruction and the method’s robustness against noise and missing points.To mitigate this trade-off,we developed a leaf surface reconstruction method that reduces the effects of noise and missing points while maintaining surface reconstruction accuracy by capturing two components of the leaf(the shape and distortion of that shape)separately using leaf-specific properties.This separation simplifies leaf surface reconstruction compared with conventional methods while increasing the robustness against noise and missing points.To evaluate the proposed method,we reconstructed the leaf surfaces from 3D point clouds of leaves acquired from two crop species(soybean and sugar beet)and compared the results with those of conventional methods.The result showed that the proposed method robustly reconstructed the leaf surfaces,despite the noise and missing points for two different leaf shapes.To evaluate the stability of the leaf surface reconstructions,we also calculated the leaf surface areas for 14 consecutive days of the target leaves.The result derived from the proposed method showed less variation of values and fewer outliers compared with the conventional methods.

Estimates of Maize Plant Density from UAV RGB Images Using Faster-RCNN Detection Model:Impact of the Spatial Resolution

Early-stage plant density is an essential trait that determines the fate of a genotype under given environmental conditions and management practices.The use of RGB images taken from UAVs may replace the traditional visual counting in fields with improved throughput,accuracy,and access to plant localization.However,high-resolution images are required to detect the small plants present at the early stages.This study explores the impact of image ground sampling distance(GSD)on the performances of maize plant detection at three-to-five leaves stage using Faster-RCNN object detection algorithm.Data collected at high resolution(GSD≈0:3 cm)over six contrasted sites were used for model training.Two additional sites with images acquired both at high and low(GSD≈0:6 cm)resolutions were used to evaluate the model performances.Results show that Faster-RCNN achieved very good plant detection and counting(rRMSE=0:08)performances when native high-resolution images are used both for training and validation.Similarly,good performances were observed(rRMSE=0:11)when the model is trained over synthetic low-resolution images obtained by downsampling the native training high-resolution images and applied to the synthetic low-resolution validation images.Conversely,poor performances are obtained when the model is trained on a given spatial resolution and applied to another spatial resolution.Training on a mix of high-and low-resolution images allows to get very good performances on the native high-resolution(rRMSE=0:06)and synthetic low-resolution(rRMSE=0:10)images.However,very low performances are still observed over the native low-resolution images(rRMSE=0:48),mainly due to the poor quality of the native low-resolution images.Finally,an advanced super resolution method based on GAN(generative adversarial network)that introduces additional textural information derived from the native high-resolution images was applied to the native low-resolution validation images.Results show some significant improvement(rRMSE=0:22)compared to bicubic upsampling approach,while still far below the performances achieved over the native high-resolution images.

Easy MPE:Extraction of Quality Microplot Images for UAV-Based High-Throughput Field Phenotyping

Microplot extraction(PE)is a necessary image processing step in unmanned aerial vehicle-(UAV-)based research on breeding fields.At present,it is manually using ArcGIS,QGIS,or other GIS-based software,but achieving the desired accuracy is timeconsuming.We therefore developed an intuitive,easy-to-use semiautomatic program for MPE called Easy MPE to enable researchers and others to access reliable plot data UAV images of whole fields under variable field conditions.The program uses four major steps:(1)binary segmentation,(2)microplot extraction,(3)production of∗.shp files to enable further file manipulation,and(4)projection of individual microplots generated from the orthomosaic back onto the raw aerial UAV images to preserve the image quality.Crop rows were successfully identified in all trial fields.The performance of the proposed method was evaluated by calculating the intersection-over-union(IOU)ratio between microplots determined manually and by Easy MPE:the average IOU(±SD)of all trials was 91%(±3).

Identifying potential field sites for production of cellulosic energy plants in Asia

Cellulosic bioethanol produced from non-edible plants avoids food-fuel competition.Growing such plants on marginal non-arable lands also avoids the use of farmland.In this study,attempts were made to identify potential field sites for cellulosic bioethanol production in Asia.In this study,GIS databases containing information about requirements such as land use,landform,and climate were superimposed.Areas with terrestrial constraints were then removed from the candidate field sites using a terrain slope database.The remaining lands were evaluated using a net primary production(NPP)database.Of these areas,southern and eastern India,northeastern Thailand,and southern Sumatra(Indonesia)had high NPP.In the 2nd phase,local information regarding infrastructure,and agriculture were analyzed.Field-establishment feasibility was high for eastern India and southern Sumatra.Potential field sites were then located in satellite images of these two areas.In the 3rd phase,soils around potential sites were evaluated.Local residents were interviewed to estimate the cost of producing plants for biomass energy.Sites selected using this simple method are suitable for biomass production.