Kathleen Brown of Penn State (@penn_state): elaboration of “steep deep and cheap”: data from multiple species (rice, maize, beans etc), environments (various stress conditions) and multiple levels of root phenotypes (anatomical traits, shovelomics and 3D-reconstruction). 

Michael A. Beck from Univ. of Winnipeg (@uwinnipeg): Progress towards a plug-and-play system for generating large numbers of labeled images in a robust and high-throughput fashion for training dataset to facilitate the application of machine learning in plant science. 

Ursula Ruiz-Vera (@URuizvera) from UIUC: Application of ground penetrating radar (GPR) in non-destructively monitoring the root vegetable Cassava under simulated CO2-enriched condition in field over multiple growing stages improves estimation for biomass. 

Heidi Dungey (@DungeyHeidi) from Scion: Highlights of the recent launched international Forest Phenotyping Working Group (FPWG) meeting. Also showcase of using LIDAR for area-based and individual tree-based phenotyping for precision forestry.

Elizabeth Castillo of Donald Danforth Center (@DanforthCenter): Time-lapse sideview imaging of Setaria viridis enabled by @Raspberry_Pi cameras under different temperatures/light intensities. Analyses using @plantcv reveal how different organs of plant respond to the environmental factors.

Lightening talks:

Haichao Guo (@haichao_guo) from the @LarryMattYork lab: An integrated platform and high-throughput pipeline for measuring specific root respiration (SRR) of wheat seedling. Genetic mapping using a winter wheat diversity panel is on the way.

Mon-Ray Shao from the @RootDevo lab: Phenotyping 3D-architecture of panicles and >1,000 root crowns of sorghum association panel (SAP) by combining X-ray computed tomography (XRT) with detailed morphometrics.

Sara Tirado of @SpringerNathan lab: Decomposing and utilizing the substantial phenotypic variation from top-down hyperspectral imaging of maize seedling (that mimics top-down view of maize field settings) for classification of genotypes and abiotic stress conditions.

Tae-Kyeong Noh from Seoul National University (@SNUnow): Evaluating herbicidal activities of various herbicides with different modes of action by analyzing parameters extracted from spectral images of herbicide-treated early-stage weed plants. 

Maarten Ameye from Ghent University (@ugent): Mechanisms underlying Fusarium Head Blight (FHB) unveiled by coupling gene expression data and monitoring the interaction of F. graminearum and F. poae on wheat leaves and ears enabled by PathoViewer system.

James Kim from @USDA_ARS: Suites of customized softwares were developed to process field data and images to extract plot-level metrics of plant phenotypes. Features including GUI, data visualization, batch processing, GIS interface, geometric/spectral calibration. 

Vasit Sagan of Saint Louis Univ. (@SLUAandS): Complementing satellite remote sensing data with low-cost sensors integrated on UAV provides better monitoring of field crops and improved accuracy for yield prediction.

Sina Roth of @Bayer4Crops: Bayer’s phenotyping greenhouse system in Frankfurt. Many data preprocessing needed (true for all!) before phenotypic data extraction: 90% of input data were removed!

Reisha Peters (@reisha_peters) of University of Saskatchewan (@usask): Expanding the application of biochemical leaf models by taking leaf biophysical properties (surface roughness, waxes etc.) into account. New models highlighted the importance of leaf surface phenotypes in spectral modeling.

Tim Beissinger of University of Göttingen (@uniGoettingen): Pipeline coupling single-plant GWAS (spGWAS) with BSA taking advantages of replication at the level of alleles within a population instead of replicated individuals, showcased by genetic mapping using the Shoepeg maize population.

Kate Seegmiller from University of Idaho (@uidaho): The top part of maize stalk is morphologically optimized for wind loading. 945 dried stalks from 82 hybrids were subjected to both three-bending test and RPR and were analyzed to determine the theoretical loading profile along stalk.

Matt Maimaitiyiming from @EnologyLab: Field-based hyperspectral imaging and PLSR model provide alternatives for traditional labor-intensive way to acquire key physiological parameters for grapevine in early stages.

Jaebum Park of @USDA_ARS: Image-based potato tuber size measurement (IPTSM) system: a high-throughput and accurate way to measure potato tuber size and shape.

Nicholas Pugh of the University of Arizona (@uarizona): Optimizing the placement and number of ground control points (GCPs) improves non-RTK (real-time kinematic) UAV’s accuracy of projects significantly, while the improvement for RTK-UAV is marginal.

Blanche Leyeza of University of Saskatchewan(@usask): Generating and utilizing(large volume of) synthetic images from real microplot images of Brassica Carinata for machine learning model training results in higher correlation with ground truth and improved seedling emergence rate accuracy.

Daniel Pflugfelder of Jülich Research Centre (@fz_juelich): Identify root traits for early wheat seedling in controlled condition (but real soil) using root MRI in Juelich. Traits including total length over time, temporal differences in emergence, angles and growth strategies.

Nick Lauter of @USDA_ARS and Iowa State University (@iastate_cals): Genetic pathways underling maize epidermis revealed by GWAS, transcriptomics analysis and more importantly, phenotypic data collected by HTP and high-res photomicroscopy from >10,000 maize leaves!

Carolyn Lawrence-Dill (@IAcornflake) of Iowa State University (@iastate_cals) and Philipp von Gillhaussen from Universität Bayreuth (@unibt): Why the International Plant Phenotyping Network (IPPN) and other PPNs are awesome 

Koushik Nagasubramanian (@koushik_here) of Iowa State University (@ISU_ECpE): Reducing the number of labeled data needed by choosing smartly: an empirical analysis of deep active learning methods for image-based plant phenotyping

Amir Ahkami (@Amir_Ahkami) of Pacific Northwest National Laboratory (@PNNLab): Monitoring photosynthesis-specific parameters of C3 and C4 model grasses with image-based Dynamic Environmental Photosynthesis Imager (DEPI), coupling multi-moics data provides insights into plant’s responses to CO2 elevation.

Samantha Neeno (@samiSpectral) of Purdue University (@LifeAtPurdue): Quantification of within-organ hyperspectral data variability in soybean under abiotic treatments (drought/N) provides useful information to remote sensing stress-detection models.

Camilo Valdes from Florida International University (@FIU): The application of a CNN framework (autoencoder) in non-destructive 3D-CT (computed tomography) root image segmentation, showcased with multiple species including cassava, potato, bean, and maize.

Yu Jiang (@Yu_Cosmo_Jiang) of Cornell (@CornellAgriTech): Application of a deep learning-based object detection model to evaluate fungicide effectiveness by identifying and quantifying the
severeness of downy mildew in grape.

Maya Kleiman of Agricultural Research Organization in Israel: Understanding plant surfaces and microorganisms’ interactions using Biomimetics, showcased by the examples of soft rot on calla and botrytis on tomato.

Even possible to synthesize the surface of roots and be used to study root knot nematode interactions!

Michael Kudenov of North Carolina State U (@NCState): Progress towards developing an emerging method of performing BRDF (bidirectional reflectance distribution function) corrections using imaging polarimetry for hyperspectral imagery with HTP scenarios.