基于改进SSD的果园行人实时检测方法

农田障碍物的精确识别是无人农业车辆必不可少的关键技术之一。针对果园环境复杂难以准确检测出障碍物信息的问题,提出了一种改进单次多重检测器(Single shot multibox detector,SSD)深度学习目标检测方法,对田间障碍物中的行人进行识别。使用轻量化网络MobileNetV2作为SSD模型中的基础网络,以减少提取图像特征过程中所花费的时间及运算量,辅助网络层以反向残差结构结合空洞卷积作为基础结构进行位置预测,在综合多尺度特征的同时避免下采样操作带来的信息损失,基于Tensorflow深度学习框架,在卡耐基梅隆大学国家机器人工程中心的果园行人检测开放数据集上进行不同运动状态(运动、静止)、不同姿态(正常、非正常)和不同目标面积(大、中、小)的田间行人识别精度和识别速度的对比试验。试验表明,当IOU阀值为0. 4时,改进的SSD模型田间行人检测模型的平均准确率和召回率分别达到了97. 46%和91. 65%,高于改进前SSD模型的96. 87%和88. 51%,并且参数量减少至原来的1/7,检测速度提高了187. 5%,检测速度为62. 50帧/s,模型具有较好的鲁棒性,可以较好地实现田间环境下行人的检测,为无人农机的避障决策提供依据。     

气相色谱内标法测定乳油中的受控溶剂

本文以毛细管气相色谱内标法测定乳油中的甲醇、甲苯、二甲苯、N,N-二甲基甲酰胺和萘5种有害溶剂。采用Stabilwax-DA毛细管色谱柱(30m×0.32mm,0.25μm)及氢火焰离子检测器,以正癸烷为内标物计算含量。测定结果不受样品中有效成分的干扰,5种目标物的线性关系良好,样品3个水平的加标回收率在98.72%~102.70%之间,相对标准偏差3.0%。对市售25种乳油产品的实际检测应用表明,该方法简便快速,准确可靠。     

枯草芽孢杆菌随水滴施防治棉花黄萎病的应用研究

通过小区试验与田间示范相结合,对枯草芽孢杆菌随水滴施防治黄萎病的效果以及对棉花产量的影响展开研究。结果表明:在小区试验中,枯草芽孢杆菌随水滴施后,对黄萎病具有一定的防治效果,防效随试验施药量增加而增加;同时较对照病指明显降低,尤其是666.7m2用量600g处理。2年中病指变化不明显,在产量上,施药处理明显高于对照处理,其中处理3增产最为明显,2年平均增产达12.78%。大田示范中,防治黄萎病效果同样明显,2年平均病指低于对照6.0以上,产量增加12.99%以上。所以,小区与田间示范均证明枯草芽孢杆菌随水滴施可以防治黄萎病发生,并能增加棉花产量。     

Dual action of Catsia tora in tea plantations: repellent volatiles and augmented natural enemy population provide control of tea green leafhopper

The tea green leafhopper Empoasca vitis Göthe (Homoptera: Cicadellidae) has become the key insect pest affecting tea plantations in southern China. The control of E. vitis with chemical insecticides is not only costly but also has serious undesirable effects. It is therefore imperative that environmentally sound and sustainable management alternatives be developed. The aim of this study was to evaluate the potential of using the non-host plant Catsia tora (Fabaceae) as a cover crop for the control of E. vitis in tea plantations. The volatile organic compounds derived from the C. tora plants were shown to be repellent to leafhoppers. These compounds were collected through headspace absorption, and its 19 components were identified through gas chromatography–mass spectrometry. Based on the behavioral responses of E. vitis to 11 of these volatile compounds (those with relative amounts greater than 1), three bioactive compounds (p-cymene, limonene, and 1,8-cineole) were observed to play a key role in the regulation of the E. vitis population and were significantly effective at repelling leafhoppers. The intercropping of C. tora in a tea field markedly reduced the E. vitis population levels and promoted an increase in the natural enemies of this pest, such as spiders, coccinellids, and lacewings. Our findings suggest that the sustainable pest management achieved using C. tora as a cover crop offers an interesting alternative strategy for the control of this important tea pest.     

基于土地开发整理修订农用地分等结果的一种方法

土地开发整理是实现耕地占补平衡的重要手段,补充耕地质量问题也一直是耕地占补平衡工作中的薄弱环节。针对补充耕地质量问题,以及已取得的农用地分等成果,考虑土地利用数据共享的原则,对在土地开发整理项目耕地等级评定中如何充分和合理利用农用地分等成果进行探讨,即对土地开发整理项目的耕地等级评定过程中,把土地开发整理工程作用的因素作为补充指标对耕地等级指数进行修订。在充分考虑补充耕地质量的基础上,依据土地开发整理工程内涵,建立了适宜土地开发整理项目中耕地等级评定的一种技术方法。并通过大石桥市虎庄镇土地开发整理项目的实证,取得令人满意的效果。     

马铃薯青枯病抗性鉴定新方法

采用传统方法和改进方法(取名为茎枝菌液浸泡法)对马铃薯的青枯病抗性进行了鉴定和评价,其结果是一致的,但茎枝菌液浸泡法更加简便快速、经济有效、可操作性和可重复性更好,并且在番茄的青枯病抗性鉴定上也得到了验证。该方法对于茄科作物的种质资源抗病性评价和筛选,特别是对于病菌诱导产生抗性的抗(感)病植株的快速鉴定及其幼苗的先期快速筛选具有重要意义,可望成为一种理想的茄科作物室内抗病鉴定方法。     

人参中人参多糖检测方法研究

目的测定人参中人参多糖的含量,并建立人参多糖的含量测定方法。方法采用紫外分光光度法,苯酚-浓硫酸比色法于波长486nm处测定吸光度。结果人参多糖含量为13.31g/100g,标准曲线的线性范围为0.9991~0.9998,平均加样回收率为98.2%,RSD=2.23%。结论苯酚-浓硫酸比色法测定人参中多糖含量,方法简便、准确、稳定性好。     

品种资源现状综述

本文概述了种质资源的重要性,阐述了我国和西藏种质资源研究的现状,针对西藏农作物种质资源研究中存在的主要问题,提出了西藏农作物种质资源研究的建议与设想。     

农户种植业结构优化与农牧结合技术方法

本文以13户农户为研究对象,通过新品种引进,开展农作物高产创建,实行农牧结合,优化了种植业结构,为种植业结构调整和农牧结合提供了技术经验与建议。     

Variation analysis and sample size estimation for growth indicators during PIT-tag-assisted family construction of channel catfish (Ictalurus punctatus)

Passive integrated transponder (PIT) tags can be used to overcome environmental variations caused by individual fish tanks during family selection of the channel catfish (Ictalurus punctatus), which thus permits growth comparisons among different families under identical environmental conditions. Power analysis was used to analyse variations in various growth performance measures during PIT-tag-assisted family construction of channel catfish, and this allowed sample sizes to be estimated. The experimental data suggested that, for a particular fish species, when having a low variation coefficient of fish body weight at the time of PIT marking and maintaining consistent individual sizes, the sample size required during a family growth experiment could be effectively reduced. This present study lays a theoretical foundation for analyzing variations in growth performance indicators and estimating the sample sizes during PIT-tag-assisted family construction of fish experiments.