Welcome to Smart Agriculture

Table of Content

    30 April 2019, Volume 1 Issue 2
    Invited Article
    Advances in diagnosis of crop diseases, pests and weeds by UAV remote sensing | Open Access
    Lan Yubin, Deng Xiaoling, Zeng Guoliang
    2019, 1(2):  1-19.  doi:10.12133/j.smartag.2019.1.2.201904-SA003
    Asbtract ( 4938 )   HTML ( 6146)   PDF (1165KB) ( 7002 )  
    References | Related Articles | Metrics

    Rapid acquisition and analysis of crop information is the precondition and basis for carrying out precision agricultural practice. Variable spraying and agricultural operation management based on the actual degree of crop diseases, pests and weeds can reduce the cost of agricultural production, optimize crop cultivation, improve crop yield and quality, and thus achieve precise agricultural management. In recent years, with the rapid development of UAV industry, UAV agricultural remote sensing technologies have played an important role in monitoring crop diseases, insects and weeds because of high spatial resolution, strong timeliness and low cost. Firstly, this research introduces the basic idea and system composition of precision agricultural aviation, and the status of UAV remote sensing in precision agricultural aviation. Then, the common UAV remote sensing imaging and interpreting methods were discussed, and the progress of UAV agricultural remote sensing technologies in detecting crop diseases, pests and weeds were respectively expounded. Finally, the challenges in the development of UAV agricultural remote sensing technologies nowadays were summarized, and the future development directions of UAV agricultural remote sensing were prospected. This research can provide theoretical references and technical supports for the development of UAV remote sensing technology in the field of precision agricultural aviation.

    Overview Article
    Key technology analysis and research progress of UAV intelligent plant protection | Open Access
    Xu Min, Zhang Ruirui, Chen Liping, Tang Qing, Xu Gang
    2019, 1(2):  20-33.  doi:10.12133/j.smartag.2019.1.2.201812-SA025
    Asbtract ( 2887 )   HTML ( 2452)   PDF (1405KB) ( 3008 )  
    References | Related Articles | Metrics

    UAV plant protection operation faces very complicated environmental conditions. On one hand, its ultra low altitude operations are vulnerable to ground structures and basic hydropower facilities; on the other hand, the effectiveness of plant protection operation is strong, and it is necessary to spray the pesticides to the specific parts of crops at the prescribed time so as to ensure good pesticide application effect. At present, UAV plant protection technology mainly refers to the existing mature technology and flight platform in general aviation field to basically "fly and spray". However, the lack of penetrating research and theoretical guidance on environmental perception in farmland operation, the movement mechanism of droplets under the rotor airflow, and the penetrability of the droplet to different crops canopy lead to low penetration rate of the UAV plant protection operation, easy drifting, frequent accidents, large damage probability and low comprehensive operational efficiency. Benefiting from the breakthroughs in artificial intelligence, parallel computing technology and intelligent hardware, the UAV plant protection technology is developing in the direction of intellectualization, systematization and precision. The real-time perception of the environment under non established conditions, intelligent job decision method based on intelligent recognition of crop diseases and pests, the control of the toward-target pesticide spraying control based on the variable of wind field droplet deposition model and the data based job evaluation system have gradually become the key technology of the UAV intelligent plant protection. The manuscript analyzed and summarized the research status and technical achievements in the field of UAV intelligent plant protection from the field information perception, the modeling and optimization control of accurate pesticide application, the evaluation and monitoring of the operation effect. Based on the existing research, the research also predicted the development trend of the key technologies of intelligent UAV plant protection in the future. The clustering method of hyper-spectral image acquisition and computational intelligence based deep learning recognition will become the key technology for real-time and efficient acquisition of crop target information in plant protection work, which greatly improves the accuracy of remote sensing information inversion recognition; machine vision and multi machine cooperative sensing technology can acquire dynamic information of field operation at multiple levels and time; the high precision droplet spectrum control technology independently controlled by nozzle design and the precision variable spraying control technology based on the wind field model can further improve the droplet deposition effect and reduce the liquid drifting; the breakthrough of high accuracy mesh solution technology will change the prediction mode of droplet drift from artificial experience judgment to computer simulation and numerical deduction; the job path planning technology will greatly improve the efficiency of multi machine and multi area operation and reduce the distance of invalid operation; the job quality evaluation based on the real-time data of the sensor and the operation supervision system of large data technology will replace people to effectively control the process of the UAV plant protection operation, achieve data and transparency of plant protection, and ensure the process is observable and controllable.

    Information Perception and Acquisition
    Corn plant disease recognition based on migration learning and convolutional neural network | Open Access
    Chen Guifen, Zhao Shan, Cao Liying, Fu Siwei, Zhou Jiaxin
    2019, 1(2):  34-44.  doi:10.12133/j.smartag.2019.1.2.201812-SA007
    Asbtract ( 2265 )   HTML ( 706)   PDF (4817KB) ( 2440 )  
    References | Related Articles | Metrics

    Corn is one of the most important food crops in China, and the occurrence of disease will result in serious yield reduction. Therefore, the diagnosis and treatment of corn disease is an important link in corn production. Under the background of big data, massive image data are generated. The traditional image recognition method has a low accuracy in identifying corn plant diseases, which is far from meeting the needs. With the development of artificial intelligence and deep learning, convolutional neural network, as a common algorithm in deep learning, is widely used to deal with machine vision problems. It can automatically identify and extract image features. However, in image classification, CNN still has problems such as small sample size, high sample similarity and long training convergence time. CNN has the limitations of expression ability and lack of feedback mechanism, and data enhancement and transfer learning can solve the corresponding problems. Therefore, this research proposed an optimization algorithm for corn plant disease recognition based on the convolution neural network recognition model combining data enhancement and transfer learning. Firstly, the algorithm preprocessed the data through the data enhancement method to expand the data set, so as to improve the generalization and accuracy of the model. Then, the CNN model based on transfer learning was constructed. The Inception V3 model was adopted through transfer learning to extract the image characteristics of the disease while keeping the parameters unchanged. In this way, the training process of the convolutional neural network was accelerated and the over-fitting degree of the network was reduced. The extracted image features were used as input of the CNN to train the network, and finally the recognition results were obtained. Finally, the model was applied to the pictures of corn diseases collected from the farmland to accurately identify five kinds of corn diseases. Identification test results showed that using data to enhance the CNN optimization algorithm and the migration study on the average recognition accuracy main diseases of com (spot, southern leaf blight, gray leaf spot, smut, gall smut) reached 96.6%, which compared with single CNN, has greatly improved the precision and identification precision by 25.6% on average. The average processing time of each image was 0.28 s, shortens nearly 10 times than a single convolution neural network. The experimental results show that the algorithm is more accurate and faster than the traditional CNN, which provides a new method for identification of corn plant diseases.

    Recognition and localization method of occluded apples based on K-means clustering segmentation algorithm and convex hull theory | Open Access
    Jiang Mei, Sun Sashuang, He Dongjian, Song Huaibo
    2019, 1(2):  45-54.  doi:10.12133/j.smartag.2019.1.2.201903-SA003
    Asbtract ( 1200 )   HTML ( 445)   PDF (1240KB) ( 1211 )  
    References | Related Articles | Metrics

    Accurate segmentation and localization of apple objects in natural scenes is an important part of wisdom agriculture research for information perception and acquisition. In order to solve the problem that apples recognition and positioning are susceptible to occlusion of leaves in natural scenes, based on the K-means clustering segmentation algorithm, the object recognition algorithm based on convex hull theory was proposed. And the algorithm was compared with the object recognition algorithm based on removing false contours and the full-contour points fitting object recognition algorithm. The object recognition algorithm based on convex hull theory utilized that apples were like circle, combining K-means algorithm with Otsu algorithm to separate fruit from background. The convex polygon was obtained by convex hull theory and fit it circle to determine the position of the fruit. To verify the effectiveness of the algorithm, 157 apple images in natural scenes were tested. The average overlap rates of the object recognition algorithm based on convex hull theory, the object recognition algorithm based on removing false contour points and the full-contour points fitting object recognition algorithm were 83.7%, 79.5% and 70.3% respectively, the average false positive rates were 2.9%, 1.7% and 1.2% respectively, and the average false negative rates were 16.3%, 20.5% and 29.7% respectively. The experimental results showed that the object recognition algorithm based on convex hull theory had better localization performance and environmental adaptability compared to the other two algorithms and had no recognition error, which can provide reference for occluded fruits segmentation and localization in the natural scenes.

    Information Processing and Decision Making
    Energy optimization strategy for wireless sensor networks in large-scale farmland habitat monitoring | Open Access
    Zhang Xiaohan, Yin Changchuan, Wu Huarui
    2019, 1(2):  55-63.  doi:10.12133/j.smartag.2019.1.2.201812-SA024
    Asbtract ( 1016 )   HTML ( 314)   PDF (1207KB) ( 1098 )  
    References | Related Articles | Metrics

    Wireless sensor network (WSN) has been widely deployed in precision agriculture to improve the crop production. However, we still face many challenges for large-scale farmland habitat monitoring, the energy shortage for battery-powered sensor nodes, the complicated propagation environment for wireless signals during different growing-up periods of the crops, the optimization of the coverage of the sensor nodes, etc. In order to guarantee the holeless coverage of the sensor nodes during the whole life of the crops, the WSNs are typically deployed with high density. Therefore, some of the sensors in the network are redundant in certain growing-up periods of the crops. And also the data collected by each sensor node may have strong temporal correlation. Recently, compressive sensing (CS) has received much attention due to its capability of reconstructing sparse signals with the number of measurements much lower than that of the Nyquist sampling rate. With the rapid progress of CS based sparse representation, matrix completion (MC) theory was proposed very recently. According to the MC theory, a low-rank matrix can be accurately rebuilt with a few number of entries in the matrix. Matrix completion provides the advantage of sampling small set of data at sensor nodes without requiring excessive computational and traffic loads, which meets the requirement of energy-efficient data gathering and transmitting in WSNs. In this research, by considering the characteristics that the sensor nodes are redundant in some growing-up periods of the crops and the data collected by sensor-nodes usually share a strong spatial and temporal correlation among them in WSNs for large-scale farmland habitat monitoring, we put forward a MC based two-step energy saving optimization algorithm to reduce both the energy consumption of the data acquisition and data transmission process in WSNs and achieved the purpose of prolonging the network lifetime. Firstly, through the measurement of the sensor node's data information, we found the non-redundant nodes by considering the spatial correlation of the data from the sensor nodes. We would close the data acquisition units of the remaining redundant nodes and make them only transmit data as relay nodes. Secondly, we took advantage of the partial sampling scheme in matrix completion to further reduce the quantity of data. Thereby, we could reduce the energy consumption on both data collection and transmission process of the wireless sensor network. The experiment results show that the proposed algorithm reduces 83% working nodes in the network, and therefore reducing the energy cost of the network.

    Intelligent Management and Control
    Framework and recommendation for constructing the SAGI digital agriculture system | Open Access
    Wu Wenbin, Shi Yun, Zhou Qingbo, Yang Peng, Liu Haiqi, Wang Fei, Liu Jia, Wang Limin, Zhang Baohui
    2019, 1(2):  64-72.  doi:10.12133/j.smartag.2019.1.2.201812-SA021
    Asbtract ( 1503 )   HTML ( 1331)   PDF (665KB) ( 2425 )  
    References | Related Articles | Metrics

    The human society is entering the era of big data and data is becoming one of the key production elements. It is thus critical to develop the China's data-driving digital agriculture system, which would greatly promote the construction of digital China, stimulate the agriculture high-quality development and improve the agricultural competitiveness at the global market. To achieve this goal, strong integration of information is needed from multi-sources, multi-sensors, and multi-scales. This research, from the perspective of agricultural information science, describes the new framework of satellite, aerial, and ground integrated (SAGI) digital agriculture system for comprehensive agricultural monitoring, modeling, and management. The SAGI system differs from traditional digital agriculture systems and includes 5 important functionalities which are resource survey, production controlling, disaster monitoring, market early-warning and decision supporting. To make the system running in operation, it is necessary to first build an observation system, which integrates the satellite, aerial, and ground in-situ observation systems to capture more sophisticated, accurate and reliable data at different scales. The system is extremely needed for China, a large country with a great geographic difference, diverse agricultural cultivation and multiple agricultural traditions. This observing system helps to form the agricultural big data for subsequent data analysis and data mining. Secondly, using the big data collected, 4 key digitalization and monitoring tasks targeting at resource property right, production process, natural disaster and market status should be implemented so as to transform the data to knowledge. In this process, some diagnosis algorithms and models are developed to understand the growth and health of crops and animals, as well as their interaction with the agro-environment. With the above support, a management system covering the full range of agricultural production, processing, selling, management and services should be established to provide the rapid and reliable information support to decision-making as well to the local farming management, thereby guaranteeing agricultural sustainability and national food security. Thirdly, some key fields for future science and technology innovation to support the applications of the SAGI system should to be enhanced such as the standardization designing, innovation in technologies and instruments, system integration and platform development. Finally, considering the complicated and integrative characteristics of this SAG system, this research also proposed some recommendations such as holistic planning, science-technology innovation, resource sharing, multi-stakeholders participation, and expansion of application fields, so as to drive this idea to the reality.

    Regionalization research of summer corn planting in North China Plain based on multi-source data | Open Access
    Diao Xingliang, Yang Zaijie, Li Qifeng, Yu Jingxin, Zheng Wengang, Shi Leigang
    2019, 1(2):  73-84.  doi:10.12133/j.smartag.2019.1.2.201901-SA002
    Asbtract ( 1224 )   HTML ( 444)   PDF (1101KB) ( 1186 )  
    References | Related Articles | Metrics

    Accurate identification of agricultural production environment information and agricultural production characteristics, comprehensive classification of meteorological, soil and crop multi-source data, are the bases for improving the efficiency of agricultural resource utilization and optimizing the structure of agricultural cultivation. Based on the meteorological data of nearly 20 years and the statistics of com yield, this study first constructed a database of spatial and temporal distribution characteristics of climate resources and com production in North China Plain, and there were significant spatiotemporal changes in rainfall, activity accumulated temperature, sunshine hours, solar radiation and corn yield. By using the method of fine crop planting regionalization, the summer com planting areas in the North China plain were divided into 5 categories: the extremely unsuitable area, the unsuitable area, the less suitable area, the suitable area, and the most suitable area, the proportions of each type of area in the total area is about 10%, 11%, 25%, 30%, 24%, respectively, further through using the Environmental Category attribution analysis method, each large class was divided into 5 subcategories, the probability was greater than 75% the relatively stable region accounts for about 63% of the total area, the fluctuation area of less than 75% is about the stable spatial and temporal distribution of 37%; the extremely unsuitable area, the unsuitable area and the less suitable area, these three kinds of spatial and temporal distributions were relatively stable, the belonging degree was 100%, accounting for 87.67%, 70.41% and 84.28%, respectively, the fluctuation zone mainly occurs between the extremely suitable zone and the suitable zone, and between the suitable zone and the relatively suitable zone. The fine zoning of summer com in North China Plain has important guiding significance for improving the utilization efficiency of local resources and optimizing the layout of com industry.

    Intelligent Equipment and Systems
    Development of an automatic steering test bench for tractors | Open Access
    Du Juan, Li Min, Jin Chengqian, Yin Xiang
    2019, 1(2):  85-93.  doi:10.12133/j.smartag.2019.1.2.201903-SA002
    Asbtract ( 1114 )   HTML ( 334)   PDF (1055KB) ( 1494 )  
    References | Related Articles | Metrics

    In recent years, the technology of automatic tractor navigation based on satellite positioning has greatly improved the efficiency and accuracy of tractor field work. Automatic steering contributes is one of the key technologies to realize the automation and intelligentization of agricultural mechanization. It costs much time to install and test automatic steering systems for tractors in the field due to complicated conditions. An automatic steering test bench was developed to reduce time consumed in the field by conducting simulation tests on accuracy and reliability. The developed automatic steering system can be applied to the tractor after obtaining satisfactory results on the test bench, which will greatly shorten the development cycle and improve the precision of the system. In this study, a 120-horsepower tractor front axle was selected. Through the design and calculation of mechanical structure, hydraulic system and electrical control system, the tractor automatic steering test bench was built. The mechanical body consist of a tractor front axle assembly, a loading device and a mechanical frame. The hydraulic interface was reserved for post-installation of automatic steering devices. Two inertial measurement units were used to test the steering system performance by recording the rotation angle of front wheels and the steering wheel. The steering wheel had a steering clearance of 16.48° and an average wheel delay time of 0.14s. In the general range and small angle range, there is a small deviation between the actual corner and the theoretical corner. Responsibility and stability met requirements for agricultural machinery steering. Experimental results show that the test bench has stable performance in terms of status detection, steering control and measurement analysis, which could meet requirements for verifying working parameters of automatic steering devices. The research provides an efficient and reliable test bench for commissioning and performance testing of agricultural machinery automatic steering.