Plant Phenomics and Image Analysis (植物表型组学与图像分析)
主讲：Ji Zhou, 周济，英国BBSRC Earlham Institute，University of East Anglia & 南京农业大学表型交叉研究中心
Remote Sensing and IoT for Phenomics（遥感和物联网技术在表型研究中的应用）
主讲：Daniel Reynolds（周济实验室, 英国BBSRC Earlham Institute）
Machine Learning for Plant Phenomics (机器学习在植物表型中的应用)
主讲：Aaron Bostrom （周济实验室, 英国BBSRC Earlham Institute）
Introduction of AgriPheno Plant Phenotyping Facility and Research Project (AgriPheno植物表型平台介绍及科研项目进展)
主讲：Hong Zhang, 张弘, 虎途老虎机国际娱乐平台_虎途老虎机娱乐最新网站科技股份有限公司
Engineering Cost-effective Intelligent Phenotyping Complete Set Instrumentation/facilities for precise crop breeding (大宗作物表型筛选精准育种成套装备、仪器与系统)
● Plant Phenomics and Image Analysis (植物表型组学与图像分析)
主讲Ji Zhou, 周济，英国BBSRC Earlham Institute，University of East Anglia, & 南京农业大学表型交叉研究中心
With the maturation of high-throughput and low-cost genotyping platforms, the current bottleneck in breeding, cultivation and crop research lies in phenotyping and phenotypic analyses. Recent phenotyping technologies invented by industry and academia are capable of producing large image- and sensor-based data. However, how to effectively transform big data into biological knowledge is an immense challenge that urgently requires a cross-disciplinary effort. In the talk, I will introduce our research-based phenotyping platforms at Norwich Research Park, ranging from the sky to cells, including AirSurf (automated aerial analytic software), Phenospex (an in-field 3D laser scanning platform), CropQuant (a low-cost distributed crop monitoring system), SeedGerm (a machine-learning based seed germination device), Leaf-GP (an open-source software for quantifying growth phenotypes), and high content screening systems for cellular phenotype measurements. Through these examples, I will introduce our multi-scale phenomics solutions developed for different biological questions on bread wheat, brassica, and other plant species, including linking phenotypic analyses to the assessment of genes controlling performance-related traits, QTL analysis of yield potential, gene discovery using near isogenic lines (NILs), quantifying genotype-by-environment interactions (GxE) to assess environmental adaptation, etc. I will also talk about how to utilise open scientific and numeric libraries for data calibration, annotation, image analysis and phenotypic analyses.
● Remote Sensing and IoT for Phenomics（遥感和物联网技术在表型研究中的应用）
主讲Daniel Reynolds（周济实验室, 英国BBSRC Earlham Institute）
A high-level overview of remote sensing, Internet of Things (IoT) and how they are applied to Plant Phenomics. Latest remote sensing and IoT provide high-resolution and high-frequency environmental measurements when compared to traditional manual methods. Distributed sensor networks such as the CropQuant platform allow researchers to record the environment of in-field or indoor experiments without manual intervention, which allow the capture of dynamic environmental changes throughout key growing stages. The lecture will introduce the techniques and applications of IoT and remote sensing in plant phenomics, covering (1) what is IoT with respect to sensing networks, (2) the hardware available and suitable for IoT including digital and analogue sensors, (3) single-board computers and microcontrollers, (4) control software and interfacing with IoT devices, (5) data transmission and retrieval, and finally (6) the management of multiple devices and collation of remote data. The lecture will not cover technical details and mainly focus on the introduction of how remote sensing and IoT could be used for phenomics.
● Machine Learning for Plant Phenomics (机器学习在植物表型中的应用)
主讲 Aaron Bostrom （周济实验室, 英国BBSRC Earlham Institute）
An introduction to machine learning and how to apply it in plant phenomics. Machine learning is a tool that has been gaining attention due to many advances in the last decade. This talk aims to provide a summary of machine learning techniques, simple and intuitive explanations and demonstrations about how machine learning has been applied to different real-world problems. In particular, generalisation and how to design training datasets and experimentation with machine learning in mind will be explained. The lecture will finish with some of Aaron’s current and previous work, and where machine learning have been applied to real world problems such as our AirSurf on lettuces yield prediction as well as SeedGerm software on seed germination measurements together with industrial leaders such as G’s Growers and Syngenta.
● Engineering Cost-effective Intelligent Phenotyping Complete Set Instrumentation/facilities for precise crop breeding (大宗作物表型筛选精准育种成套装备、仪器与系统)
It plays an important role for high-throughput phenotyping in cutting-edge crop breeding field, and this automation generates heterogeneous measuring data for subsequent meta-analyses, modeling, and ground-truth dataset building. Traditional researches focus on an individual instrument or data processing algorithms. We advocate that the crop breeeding issue has to be addressed with a systematic paradigm, ranging from building cost-effective infrastructure to leveraging crowd-sourcing applications, and to process standardization.The roadmap for conducting phenotyping-based breeding is depicted as, first, plant organ-specific phenotyping parameter index sets for crop breeding are optimally determined, and corresponding phenotyping instrumentation are introduced. Second, an entity-relationship data aggregation model is built to organize and present the phenotyping big data; Third, a paradigm of creating a phenotyping database is proposed to facilitate crop breeding. Finally, a formal GPEM database for constructing a crop breeding phenotyping database is established, which highlights the plant morphometric data retrieval and data mining. This data aggregation scheme provides an effective tool and exemplary template for dealing with big plant phenotyping data acquired by different devices and equipment under user-defined resolution. The case study for creating a GPEM phenotyping database is step-by-step investigated to show the feasibility and effectiveness of plant phenotyping big-data aggregation.