Global Wheat Head Detection (GWHD) Dataset: A Large and Diverse Dataset of High-Resolution RGB-Labelled Images to Develop and Benchmark Wheat Head Detection Methods

Plant Phenomics, Journal Year: 2020, Volume and Issue: 2020

Published: Jan. 1, 2020

The detection of wheat heads in plant images is an important task for estimating pertinent traits including head population density and characteristics such as health, size, maturity stage, the presence awns. Several studies have developed methods from high-resolution RGB imagery based on machine learning algorithms. However, these generally been calibrated validated limited datasets. High variability observational conditions, genotypic differences, development stages, orientation makes a challenge computer vision. Further, possible blurring due to motion or wind overlap between dense populations make this even more complex. Through joint international collaborative effort, we built large, diverse, well-labelled dataset images, called Global Wheat Head Detection (GWHD) dataset. It contains 4700 190000 labelled collected several countries around world at different growth stages with wide range genotypes. Guidelines image acquisition, associating minimum metadata respect FAIR principles, consistent labelling are proposed when developing new GWHD publicly available http://www.global-wheat.com/and aimed benchmarking detection.

Language: Английский

---

Climate and agronomy, not genetics, underpin recent maize yield gains in favorable environments

Proceedings of the National Academy of Sciences, Journal Year: 2022, Volume and Issue: 119(4)

Published: Jan. 18, 2022

Quantitative understanding of factors driving yield increases major food crops is essential for effective prioritization research and development. Yet previous estimates had limitations in distinguishing among contributing such as changing climate new agronomic genetic technologies. Here, we distinguished the separate contribution these to advance using an extensive database collected from largest irrigated maize-production domain world located Nebraska (United States) during 2005-to-2018 period. We found that 48% gain was associated with a decadal trend, 39% improvements, and, by difference, only 13% improvement potential. The fact findings were so different most studies, which gave much-greater weight potential improvement, gives urgency need reevaluate contributions advances all help guide future investments development achieve sustainable global security. If progress also slowing other environments crops, crop-yield gains will increasingly rely on improved practices.

Language: Английский

---

WheatNet: A lightweight convolutional neural network for high-throughput image-based wheat head detection and counting

Neurocomputing, Journal Year: 2022, Volume and Issue: 489, P. 78 - 89

Published: March 14, 2022

Language: Английский

---

Fertilizers and Fertilization Strategies Mitigating Soil Factors Constraining Efficiency of Nitrogen in Plant Production

Plants, Journal Year: 2022, Volume and Issue: 11(14), P. 1855 - 1855

Published: July 15, 2022

Fertilizer Use Efficiency (FUE) is a measure of the potential an applied fertilizer to increase its impact on uptake and utilization nitrogen (N) present in soil/plant system. The productivity N depends supply those nutrients well-defined stage yield formation that are decisive for utilization. Traditionally, plant nutritional status evaluated by using chemical methods. However, nowadays, correct doses, absorption reflection solar radiation used. Fertilization efficiency can be increased not only adjusting dose plant’s requirements, but also removing all soil factors constrain nutrient their transport from root surface. Among them, compaction pH relatively easy correct. goal new formulas fertilizers availability synchronization release with demand. aim non-nitrogenous control effectiveness A wide range actions required reduce amount which pollute ecosystems adjacent fields.

Language: Английский

---

Interpretability of deep learning models for crop yield forecasting

Computers and Electronics in Agriculture, Journal Year: 2023, Volume and Issue: 206, P. 107663 - 107663

Published: Feb. 2, 2023

Machine learning models for crop yield forecasting often rely on expert-designed features or predictors. The effectiveness and interpretability of these handcrafted depends the expertise people designing them. Neural networks have ability to learn directly from input data train feature prediction steps simultaneously. In this paper, we evaluate performance neural network using MARS Crop Yield Forecasting System European Commission's Joint Research Centre. selected can handle sequential time series include long short-term memory (LSTM) recurrent 1-dimensional convolutional (1DCNN). Performance was compared with a linear trend model Gradient-Boosted Decision Trees (GBDT) model, trained hand-designed features. Feature importance scores variables were computed attribution methods analyzed by modeling agronomy experts. Results showed that LSTM perform statistically better than GBDT soft wheat in Germany similar all other case studies. addition, captured effect trend, static (e.g. elevation, soil water holding capacity) biomass well, but struggled capture impact extreme temperature moisture conditions. Our work shows potential deep automatically produce reliable forecasts, highlights challenges involving human stakeholders assessing interpretability.

Language: Английский

---

Harvesting the benefits of nutritional research to address global challenges in the 21st century

Journal of the World Aquaculture Society, Journal Year: 2023, Volume and Issue: 54(2), P. 343 - 363

Published: March 25, 2023

Abstract Over the past 20 years, substantial progress has been made in improving feeds and feeding technologies for most aquaculture species. Notable improvements feed conversion efficiency (through a better understanding of requirements improved management) ingredient sustainability increased capability to use wider range ingredients) have achieved. While advances many main species, there is still much be done defining requirements, especially species being farmed developing world. Gains are slowing developed but potential gains appreciable less There growing need more precisely prescribe required levels essential nutrients various additives diet based on age, genotype, environment, immune status deliver “precision nutrition” approach farming further diversify our options provide greater resilience, as different sources, including possible climate change impacts, becoming issue. demand biocircularity supply chains. Ultimately, what needed sustain future needs sustainable sources cost‐effective protein, some amino acid additives, omega‐3 fatty resources, minerals vitamin additives. The increasing new varied resources will ensure that food safety remains an important issue throughout Feed manufacturing evolved from simplistic exercise highly complex science with state‐of‐the‐art engineering, its application not consistent across all sectors, widespread pelleting, mash, trash fish Similarly, management also dichotomized between world, high reliance manual skilled labor whereas advanced systems increasingly reliant automated computer‐controlled systems.

Language: Английский

---

Global spatially explicit yield gap time trends reveal regions at risk of future crop yield stagnation

Nature Food, Journal Year: 2024, Volume and Issue: 5(2), P. 125 - 135

Published: Jan. 26, 2024

Yield gaps, here defined as the difference between actual and attainable yields, provide a framework for assessing opportunities to increase agricultural productivity. Previous global assessments, centred on single year, were unable identify temporal variation. Here we spatially temporally comprehensive analysis of yield gaps ten major crops from 1975 2010. have widened steadily over most areas eight annual remained static sugar cane oil palm. We developed three-category typology differentiate regions 'steady growth' in 'stalled floor' where is stagnated 'ceiling pressure' are closing. Over 60% maize area experiencing growth', contrast ∼12% rice. Rice wheat 84% 56% area, respectively, pressure'. show that correlates with subsequent stagnation, signalling risks multiple countries currently realizing gains growth.

Language: Английский

---

How changes in climate and agricultural practices influenced wheat production in Western Europe

Journal of Cereal Science, Journal Year: 2020, Volume and Issue: 93, P. 102960 - 102960

Published: March 14, 2020

Language: Английский

---

Methods of yield stability analysis in long-term field experiments. A review

Agronomy for Sustainable Development, Journal Year: 2021, Volume and Issue: 41(2)

Published: March 29, 2021

Abstract In the face of a changing climate, yield stability is becoming increasingly important for farmers and breeders. Long-term field experiments (LTEs) generate data sets that allow quantification different agronomic treatments. However, there are no commonly accepted guidelines assessing in LTEs. The large diversity options impedes comparability results reduces confidence conclusions. Here, we review provide guidance most encountered methodological issues when analysing major points recommend discuss individual sections following: researchers should (1) make quality approaches analysis from LTEs as transparent possible; (2) test deal with outliers; (3) investigate include, if present, potentially confounding factors statistical model; (4) explore need detrending data; (5) account temporal autocorrelation necessary; (6) explicit choice measures consider correlation between some measures; (7) dependence on mean yield; (8) trends stability; (9) report standard errors inference where possible. For these issues, pros cons various solutions examples illustration. We conclude to ample use linking up sets, publish data, so can be compared by other authors and, finally, impacts methods interpreting analyses. Consistent suggested recommendations may basis robust analyses subsequently design stable cropping systems better adapted climate.

Language: Английский

---

Machine learning in crop yield modelling: A powerful tool, but no surrogate for science

Agricultural and Forest Meteorology, Journal Year: 2021, Volume and Issue: 312, P. 108698 - 108698

Published: Nov. 10, 2021

Provisioning a sufficient stable source of food requires sound knowledge about current and upcoming threats to agricultural production. To that end machine learning approaches were used identify the prevailing climatic soil hydrological drivers spatial temporal yield variability four crops, comprising 40 years data each from 351 counties in Germany. Effects progress management breeding subtracted prior modelling by fitting smooth non-linear trends 95th percentiles observed data. An extensive feature selection approach was followed then most relevant predictors out large set candidate predictors, various meteorological Particular emphasis placed on studying uniqueness identified key predictors. Random Forest Support Vector Machine models yielded similar although not identical results, capturing between 50% 70% variance silage maize, winter barley, rapeseed wheat yield. Equally good performance could be achieved with different sets Thus identification reliable based outcome model study only but required expert's judgement. Relationships response often exhibited optimum curves, especially for summer air temperature precipitation. In contrast, moisture clearly proved less compared drivers. view expected climate change both excess precipitation heat effect deserve more attention as well crop modelling.

Language: Английский

---