基于计算机视觉的养殖动物计数方法研究综述

数量计量是动物养殖管理的基础工作,其结果对于动物养殖的生产效率、养殖成本管控及经济效益评估等具有重要意义。基于计算机视觉的计数方法解决了传统人工计数存在的测量误差大、耗时费力等问题,减轻了养殖人员的工作负担。本文统计分析了近十年的养殖动物视觉计数相关研究,从传统机器学习与深度学习两方面对养殖动物计数算法进行分析与讨论。此外,对水产养殖、畜禽养殖与特种动物养殖领域的养殖动物计数应用进行梳理与总结。同时,对目前公开发布的养殖动物计数数据集进行概述。最后,从数据集、应用场景、计数方法3方面分析讨论养殖动物计数研究面临的主要挑战,并对未来研究进行展望。     

田间作物高通量表型信息获取与分析技术研究进展

田间作物表型信息获取种类、数量以及信息处理与分析方法对于发现有价值的表型特性并确定其遗传因素有着重要影响,而传统的田间表型信息获取方法依赖于研究人员的人工采样测量,不但费时费力,还存在效率低和主观性强等缺点。为解决这一问题,田间作物高通量表型信息获取及分析技术成为了当前植物表型领域的一个研究热点。目前,表型研究主要集中在3个方面：传感器、平台和信息分析。本文从这3个方面阐述国内外田间作物高通量表型信息获取及分析技术的最新研究成果,分析表型信息获取技术中常用传感器的应用范围和使用限制条件以及不同表型信息获取平台的优缺点,总结表型信息分析的方法,提出使用时需要根据具体情况,综合考虑实际需求以及经济合理性选择和设计。最后展望田间作物表型研究未来的发展方向,将集中在多类型数据融合、数据标准化管理、多学科知识整合等方面。该项研究成果对推广田间表型信息获取技术和分析方法、促进表型研究和遗传育种研究的深入融合提供了理论参考和技术支撑。     

Assessing the potential benefits of alternative cane supply arrangements in the Australian sugar industry

Efficient production of sugar from sugarcane requires integration of growing, harvesting, transport, milling, and marketing. While research targeted to each of these sectors has added significant value in the past, further gains are limited, giving rise to an increasing focus on whole of systems research for the industry to maintain its international competitiveness. A major whole of industry issue is the need to take advantage of geographical, temporal, and crop characteristic differences in sugar yields. Through capturing most of this variability, a methodology was developed that allowed the assessment of profitability benefits from alternative harvest date of cane options throughout a mill region. This resulted in an original application of operations research. Options analyses were carried out in partnership with mill regions throughout the Australian sugar industry, showing potential gains of up to A$157/ha at a sugar price of A$250/t, with the size of the gains dependent on the international sugar price and climatic variability.     

基于物联网的浮标水质监测系统与溶解氧浓度预测模型

为促进近海养殖业信息化发展,更好地实现对近海养殖环境的监控,设计了基于浮标平台的环境监测系统。利用STM32L475微控制器定时采集光照、温度、pH值、溶解氧浓度等信息,通过物联网技术将数据传输至云监测平台,实现了多区域环境信息远程监测和多终端访问。提出了改进遗传算法BP神经网络的溶解氧浓度预测模型,实现对近海养殖环境的预测;根据所采集的数据,利用改进遗传算法对初始权重和阈值进行优化得到最优参数,在此基础上构建BP神经网络溶解氧浓度预测模型。通过试验验证了该系统海洋环境信息采集的准确性与可靠性,以及溶解氧浓度预测模型的有效性;与传统遗传算法BP神经网络预测模型相比,平均误差由0.0778mg/L降至0.0178mg/L,能够满足近海养殖的实际需求。     

畜禽养殖个体信息监测技术研究进展

畜禽养殖个体信息主要包括发情信息、分娩信息、行为信息、体重信息和健康信息。自动化、集约化养殖是养殖业的发展趋势,准确高效地监测动物个体信息有利于分析动物的生理、健康和福利状况,是实现自动化健康养殖和肉品溯源的基础。目前生产中主要依靠人工观测的方式监测动物个体信息,耗时费力且主观性强。本文阐述了畜禽养殖中发情监测、分娩监测、行为监测、体重监测和健康监测等技术的应用研究现状,展望了畜禽养殖个体监测技术后续研究方向。     

沙棘产品突围及创新路径的分析与展望

改变沙棘产品"两多、一少"(常规产品多,低劣产品多,附加值高产品少)局面,实现突围与创新,主要路径是制定系统的产品研发战略,实施持续优化的产品方案;加大研发投入,提高新产品入市标准;建设产学研合作平台,提高研发的含金量;准确的产品定位与标准化;提高产品的附加值为市场切入点;突破传统,强化沙棘"药食两用"功效,实现品种与市场创新;强化产品形象设计;打造品牌,拓展国际市场。     

基于BERT的水稻表型知识图谱实体关系抽取研究

针对水稻表型知识图谱中的实体关系抽取问题,根据植物本体论提出了一种对水稻的基因、环境、表型等表型组学实体进行关系分类的方法。首先,获取水稻表型组学数据,并进行标注和分类;随后,提取关系数据集中的词向量、位置向量及句子向量,基于双向转换编码表示模型（BERT）构建水稻表型组学关系抽取模型;最后,将BERT模型与卷积神经网络模型、分段卷积网络模型进行结果比较。结果表明,在3种关系抽取模型中,BERT模型表现更佳,精度达95.11%、F1值为95.85%。     

畜牧养殖穿戴式信息监测技术研究现状与发展分析

目前畜牧养殖环节信息监测技术依然落后,且普及率不高,无法准确、有效地掌握养殖环节农场动物本身及其生活环境的实际状况。本文在整理和总结现有穿戴式技术研究成果基础上,结合我国畜牧养殖实际情况和特点,对畜牧养殖穿戴式信息监测的工作原理、信息监测技术和穿戴式监测方式等方面进行了分析和讨论。总结得出未来研究趋势：信息获取方式由人工采集向自动化采集发展;穿戴式传感器将向微型化和柔性化方向发展;信号处理与信息传输将向多元化、复合化和智能化方向发展;信息监测方式将向系统性、整体性和自适应方向发展     

基于半监督主动学习的菊花表型分类研究

鉴于人工和专家分类模式的局限性,基于表型的菊花分类存在效率低下的问题。本文采用基于半监督主动学习技术,在已分类菊花数据的基础上,利用未标号菊花样本数据提供的信息,建立了菊花表型分类模型,提升了分类质量和效率。该模型可以不依赖外界交互,利用未标号样本来自动提升菊花分类的质量。为了训练学习模型,本文收集了菊花的表型特征数据,标注了菊花表型类别,并研究了菊花分类属性特征的编码技术。在此数据集上,采用基于图标号传播的半监督学习技术对未标号的菊花数据进行建模,为了提升半监督分类的有效性,在标号传播的基础上使用主动学习技术,采用熵最大策略来选择难以识别的样本,以改进分类质量。在该数据集上进行了试验验证,并进行了试验对比和分析,试验结果表明,本文方法能够较好地利用未标号菊花样本提升分类的精度,随着标号百分比从6.25%升至23%,识别精度达到0.7以上,标号百分比在81.25%时,平均识别精度和召回率分别达到0.91和0.88。     

Efficiency gains from water markets: Empirical analysis of Watermove in Australia

We analyse supply, demand, and welfare measures in markets where the temporary trading of water rights is reasonably active and liquid on a weekly basis. We identify four important characteristics of water trading within the Watermove program: (1) demand is highly responsive to price changes, supporting evidence that farmers’ participation is driven by the advantages water markets provide in reducing seasonal risk; (2) some efficiency enhancing trades are prevented from taking place due to physical limitations of the delivery system, thereby reducing the ability of the market to move toward equilibrium. In addition, these limitations also created price premiums in some zones; (3) variations in the types of products traded are explained by differences in the administrative charges for trading and current drought conditions; and (4) relatively large gains have been made thus far, suggesting that water markets will expand and generate additional increases in consumer and producer surplus in the future.     

Impact of Subsurface Drainage on Improvement of Crop Production and Farm Income in North-West India

The Indian Council of Agricultural research has given priority to control and manage salinity problems that have developed in north-west India. Multi-disciplinary taskforces have recommended installation of subsurface drainage for salinity control, based on design and management techniques developed by the Central Soil Salinity Research Institute (CSSRI), to rehabilitate lands with excess soil salinity. After small-scale studies, large-scale pilot projects were launched to install subsurface drainage in problem areas. One such attempt in was initiated in the north-west region of India where a large-scale drainage project was carried out with Dutch collaboration. We assessed the impact of investments in subsurface drainage in order to validate past funding on research of drainage in India. The important methods used for assessing the efficiency benefits of drainage investment were: to determine the impact of subsurface drainage in terms of net present value, internal rate of returns, consumers' surplus and producers' surplus; to assess the social welfare in terms of social equality and sustainability of the drainage system; and to examine the factors affecting the sustainability of the technology. The internal rate of return was computed to assess the efficiency parameter of subsurface drainage for salinity management. In order to measure the changes in inequality distribution of income, Gini concentration ratios were computed with and without installing sub surface drainage. The Radar Approach, a method based on a graphical display of differences between actual ideal performance, was used to quantify drainage sustainability in terms of optimizing gains and conserving, or improving the quality of soil and water resources. There were several farm-level benefits as a result of installing subsurface drainage: these included: (i) a substantial increase in farm income; (ii) cropping intensification and diversification toward high value crops; and (iii) generation employment. A high internal rate of return justified investment in subsurface drainage. Income inequalities across farms were reduced. The radar approach showed improvement in sustainability in terms of economic gains and resource conservation. Despite of these economic, social, and environmental benefits, the sustainability of subsurface drainage technology is questionable. The specific reasons include: (i) the nature of the technology; (ii) lukewarm collective action by the beneficiaries; (iii) conflicting objectives among beneficiaries; and (iv) growing numbers of free riders. To a large extent these were addressed in the study area by forming village committees. Without appropriate institutional arrangements, subsurface drainage may not yield the desired results, and in the long run may result in neglect of operation and maintenance needs and ultimately the abandonment of the technology.     

农业现代化进程中的生产函数估计*

农林牧渔业投资、农业机械化水平、化肥、塑料薄膜和农业科技水平是农业生产中重要的投入要素,将这些要素纳入扩展型C-D生产函数模型,并对2000—2019年我国农业生产经济数据进行岭回归计算,得到我国同期的农业生产函数,经变量替换和研究期调整等证实该农业生产函数具有稳健性。在我国农业现代化进程中,农林牧渔业投资、化肥施用量、塑料薄膜使用量、农业机械化水平和农业科技水平对农业总产值的产出弹性系数依次降低,分别为0.383、0.269、0.213、0.161和0.051;农业科技水平、农林牧渔业投资、塑料薄膜使用量、农业机械化水平和化肥施用量对农业总产值贡献率依次为21.392%、20.681%、15.444%、13.295%和8.684%。因此,应加大农业科技水平投入和转化、农林牧渔业投资力度和新型机械动力使用补贴等,推进农业现代化建设进程。     

Measuring the costs of African animal trypanosomosis,the potential benefits of control and returns to research

This paper addresses issues surrounding measurement of the potential productivity gains from new livestock technologies and the returns to international livestock research. The approach, applicable to many livestock production constraints and technologies, uses geographic information systems (GIS) to spatially link a biophysical herd simulation model with an economic surplus model. The particular problem examined is trypanosomosis in cattle in Africa, and the potential research product is a multi-component vaccine. The results indicate that the potential benefits of improved trypanosomosis control, in terms of meat and milk productivity alone, are $700 million per year in Africa. The disease now costs livestock producers and consumers an estimated $1340 million annually, without including indirect livestock benefits such as manure and traction. Given an adoption period of 12 years, a maximum adoption rate of 30%, a discount rate of 5%, and a 30% probability of the research being successful within 10 years, the net present value of the vaccine research is estimated to be at least $288 million, with an internal rate of return of 33%, and a benefit/cost ratio of 34:1. ©     

基于近似多位姿的轮系式钵苗移栽机构运动综合

作业轨迹形状和在特定位置移栽臂的姿态要求是移栽机构设计过程需要考虑的首要问题。为了能获得多个期望位姿且兼顾一定形状的作业轨迹,提出一种基于近似多位姿的非圆齿轮行星轮系移栽机构设计方法。基于运动学映射理论将8个期望位姿点的信息矩阵通过四元数转换至三维空间,在对该矩阵进行奇异值分解后,通过引入特征向量扩大一般解空间。结合约束条件求解得到3组RR型二杆组,根据移栽机构的工作范围选取其中两组合适的二杆组构成四杆机构,且该机构连杆上的一点可依次近似通过所有设定的位姿点形成一条完整的“8字”型轨迹。根据其中一组二杆组的运动特点求解非圆齿轮的传动比,以非圆齿轮节曲线圆度性为目标,以封闭轨迹上的部分形状可调节段拟合点位置为变量（即不改变关键轨迹段上型值点的位姿）,利用遗传算法优化得到一组具有较好圆度性的非圆齿轮节曲线。最后进行了水稻钵苗移栽机构的设计与试验,理论与试验的移栽姿态和轨迹形状具有一致性,且取苗试验的成功率达到设计要求,有效地验证了运动综合和设计方法的正确性。本研究可为具有多位姿要求的新型移栽机构的设计提供方法借鉴。     

基于辅助信息的森林蓄积量空间模拟

以北京市密云县一类清查的样地蓄积量为研究对象,结合与蓄积量相关的辅助因子,采用普通克里格法、协同克里格法对森林蓄积量进行空间插值估测,并与文献[25]同一研究区的基于偏最小二乘法回归法估测结果进行比较分析。结果表明,普通克里格法、基于辅助信息的协同克里格法、偏最小二乘回归法的蓄积量估测值与实测值间的相关系数分别为0.389、0.845、0.766;基于辅助信息的协同克里格法要优于普通克里格法和偏最小二乘回归法,能够明显提高预测精度;与普通克里格法相比,所产生的均方根误差减小了71%,预测值和实测值的相关系数提高了54%。最后生成了密云县森林蓄积量空间分布图。研究表明应用地统计学方法进行蓄积量估测具有很好的应用前景,可以为森林蓄积量的估测提供一种可行的方法。     

A bio-economic modelling project for small-scale milk production systems in South-East Brasil

This paper reports on the development and use of a bioeconomic model to simulate dairy enterprises in the Zona da Mata of Minas Gerais State in south-east Brasil.In view of the relatively small size of the units, each animal within the herd is accounted for on an individual basis and the main biological events and processes are treated stochastically. Within the limits of given herdranges for genetic potentials, actual rates of performance are estimated from the levels of feeding and other management factors. An energy balance approach is used to assess the contributions of pasture, cut forage and concentrates in the diet.The model has been used to help in identifying priority areas for applied research, to explore various combinations of herd potential and level of feeding, the behaviour of different herd sizes and the potential returns from specific technical innovations.     

Increasing animal productivity on small mixed farms in South Asia: a systems perspective

Smallholder crop–animal systems predominate in south Asia, and most of the projected future demands for ruminant meat and milk are expected to be met from the improved productivity of livestock in these mixed farming systems. Despite their importance in the sub-region, there is a paucity of information on research that incorporates animals interactively with cropping. Livestock research has tended to highlight component technologies, often treating diverse and complex mixed farming operations as a single system. Furthermore, little attention has been paid to social, economic or policy issues. Thus, many of the technological interventions have either failed to become adopted at farm level or their uptake has proved unsustainable. This paper reviews aspects of animal production in South Asia; the trends and forecasts for animal populations and products, constraints to productivity, research opportunities and some key examples of technologies that have failed to achieve their full potential on farm. A systems analysis of small-scale crop–livestock operations is advocated, as a precursor for targeting appropriate interventions at farm level to increase animal productivity and protect the natural resources base.     

基于物联网技术的企业MES系统改进研究

对现有的制造执行系统进行研究,针对企业在生产过程中出现的物流配送及装配过程实时监控能力弱、报警系统不完善和信息交互不及时等问题,提出了基于物联网技术的制造执行系统改进方案,完善了企业制造执行系统的功能,有效地实现了企业系统的集成和动态监控,对企业信息化建设具有一定的实用价值。     

Crop–animal systems in Asia: socio-economic benefits and impacts on rural livelihoods

Traditionally, small farmers in Asia have practiced mixed farming. To improve crop and animal productivity, increase farm incomes and maintain the ecological balance, several technology options have been developed through on-station and on-farm research by international organisations and national agricultural research systems. However, a review of the research reveals a paucity of information, particularly in South Asia, on the socio-economic benefits and impacts of these technologies and interventions for poor farming households. This paper presents the few case studies available which document the benefits of new technologies to improve crop–animal systems. Additionally, the paper suggests reasons for the neglect of socio-economics in these studies, and ways to strengthen this dimension.