BRENNEN

Industries / Food, farming and forestry / Agriculture and farming / Grain elevators and storage

Receiving and grading

Receiving is the work at the elevator's front door: weighing each truck, drawing a sample, measuring its moisture and quality, and grading it, so the seller is paid for what was delivered and the grain goes into a bin with grain of the same grade.

5

Milestones achieved, from 1856 to 2025.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1856

    Chicago grades its grain

    In 1856 the Chicago Board of Trade adopted uniform grading standards. Grain delivered to an elevator was inspected and dumped into a common bin kept for grain of the same grade, and the owner got a receipt for, say, a thousand bushels of No. 2 spring wheat rather than for a particular lot. The standards made those receipts accepted everywhere and freely traded, by the account of American Heritage magazine.

    americanheritage.com

    Credit

    Chicago Board of Trade

  2. 1975

    Infrared tests wheat protein

    Buyers began paying for wheat by its protein content in the mid-1960s, and the standard protein test, the Kjeldahl method of wet chemistry, took about two hours a sample. Phil Williams of the Canadian Grain Commission's Grain Research Laboratory bought a near-infrared (NIR) instrument in February 1972; NIR reads protein and moisture from the way a sample absorbs infrared light just beyond what the eye sees. In 1975 the Commission began testing samples from where grain entered the system with an automated analyzer that did 10 samples a minute, and the samples tested for protein each year rose from about 40,000 to more than 500,000, by the Manitoba Co-operator's account. In 1992 the Commission added near-infrared transmittance instruments, which read whole grain.

    manitobacooperator.cacanada.ca

    Credit

    Phil Williams, Canadian Grain Commission

  3. 2008

    Canada drops the eye test

    Western Canada kept its wheat classes apart by kernel visual distinguishability (KVD): each class had its own kernel shape and seed-coat colour, so inspectors could tell classes apart by eye from farm to ship, and a new variety could be registered only if it looked like its class. The Canadian Grain Commission (CGC) noted that KVD cannot catch a variety that looks like a class but lacks its quality, and in 2001 set up an advisory committee to find alternatives, among them variety eligibility declarations. On 11 February 2008 the agriculture minister, Gerry Ritz, announced that KVD would stop being a registration requirement for all western wheat classes from 1 August 2008, so that high-yielding wheats for feed and biofuel that look like milling wheat could be grown.

    LessonGrading by eye limits which varieties farmers can grow, and dropping it means variety must be identified some other way.

    bac-lac.wayback.archive-it.orgbac-lac.wayback.archive-it.orgbac-lac.wayback.archive-it.orgagriville.com

    Credit

    Canadian Grain Commission; Agriculture and Agri-Food Canada

  4. 2012

    One moisture method for all

    Moisture sets both the price of a load and how long it will keep. The Federal Grain Inspection Service (FGIS), the US Department of Agriculture's grain inspection agency, developed the Unified Grain Moisture Algorithm (UGMA): a meter measures the grain's dielectric constant (how strongly it stores an electric field, which rises with water content) at about 149 MHz, corrects for density and temperature, and applies one equation to every grain type. FGIS designed it so that meters from different makers give the same reading and can compete for official use. On 10 September 2012 the Perten AM 5200-A and the Dickey-john GAC 2500-UGMA replaced the Dickey-john GAC 2100 as the official meters for corn, soybeans, sorghum and sunflower, and a 2020 FGIS directive added a third model, the FOSS GAC 2500-C.

    webarchive.library.unt.eduams.usda.govgrainjournal.com

    Credit

    David Funk and FGIS; Perten Instruments; Dickey-john

  5. 2025

    Cameras join the sampler

    In Western Australia the grain handler CBH Group uses machines that photograph samples at receival and check wheat, barley and oats for weed seeds and defects that samplers used to judge by eye. After trials at 2 sites in 2023/24 and 14 in 2024/25, they ran at all Albany and Esperance zone sites in 2025/26. In eastern Australia GrainCorp has more than 40 ZoomAgri devices that check barley variety from images, and Grain Trade Australia, which writes the trade's standards, endorsed the Zoom One device for six barley varieties for the 2026/27 season, so barley can be certified and priced on its result. In January 2026 the Canadian Grain Commission reported that ZoomAgri's variety calls on 71 samples of 7 barley varieties agreed with its own DNA tests to within ±5% purity.

    cbh.com.augraincorp.com.aucanada.ca

    Credit

    CBH Group and Deimos (through their joint venture, Acova); ZoomAgri; GrainCorp; Grain Trade Australia; Canadian Grain Commission

Watch for

  1. AI grades lentils commercially

    Cropify, an Adelaide company, says its cameras and AI assess a lentil sample for defects, contamination and quality in minutes and keep a digital record of each load, in place of a grader's visual check. After an endorsement from Grain Trade Australia's Trading Standards Committee, it expected to deploy up to 25 units for the 2026 harvest; its first commercial harvest, in 2025, involved GrainCorp, Louis Dreyfus Company, ETG Wimpak and Shannon Bros. No accuracy figures have been published.

    Achieved whena receiver or standards body outside Cropify publishes how often its results match trained graders over a full harvest.

    itbrief.com.au

    Claimed by

    Cropify, August 2026

  2. Phone app grades soybeans

    Grain Discovery, Inarix and Sevita International ran a Protein Industries Canada project from March 2025 to March 2026 on a phone app that gives an AI assessment of a photographed soybean sample, mainly its variety, in under 20 seconds, with Sevita as the main pilot site. The project page says the app gives consistent and reliable grading but publishes no accuracy data.

    Achieved whena party outside the partners publishes the app's agreement with laboratory or DNA variety tests on samples it did not train on.

    proteinindustriescanada.ca

    Claimed by

    Grain Discovery, Inarix and Sevita International, June 2025

Open

  1. 01

    Graders and cameras, scored

    A public reference set of whole grain samples, for example 300 wheat and corn samples, each graded independently by at least three official inspectors and photographed under fixed lighting, so that any camera grader can be scored against the spread between human inspectors on the factors that set the grade, such as damaged kernels and foreign material.

    The brief
    Credit · unclaimed
  2. 02

    Handheld meters against the oven

    A public test of the handheld moisture meters farmers use to decide when to harvest and deliver, read side by side with the air-oven reference method and with an elevator's official-type meter on split samples of corn and soybeans, at grain temperatures from near freezing to warm.

    The brief
    Credit · unclaimed

Open 01 · the brief

Graders and cameras, scored

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public reference set of whole grain samples, for example 300 wheat and corn samples, each graded independently by at least three official inspectors and photographed under fixed lighting, so that any camera grader can be scored against the spread between human inspectors on the factors that set the grade, such as damaged kernels and foreign material.

Why it is still open

GrainSpace, a public set of 5.25 million kernel images labelled by professional inspectors, judges single kernels, not whole samples graded to a standard. The CGC's test of ZoomAgri covered barley variety only, on 71 samples, and CBH's page gives no measured accuracy for its receival machines. A trade-journal study that set image AI against five laboratory technicians on malting barley sits behind a paywall, with no data shown.

arxiv.orgcanada.cacbh.com.aubrauwelt.com

Built with what exists

A phone or camera on a copy stand with an LED panel and a colour reference card; official grading of "submitted samples", which any interested person can send to a US official inspection agency for a fee; Label Studio for labelling; an open image model such as Ultralytics YOLO, pre-trained on GrainSpace; Python for the statistics.

law.cornell.edu

  1. First step

    Collect 30 farm samples of one wheat class, split each three ways, and send the splits to two official agencies as submitted samples.

  2. Done looks like

    300 samples, each graded by three inspectors from at least two agencies, with the share of samples on which the inspectors agree on the grade published beside the share on which each tested camera system matches the inspectors' majority; images, grades and code public.

Open 02 · the brief

Handheld meters against the oven

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public test of the handheld moisture meters farmers use to decide when to harvest and deliver, read side by side with the air-oven reference method and with an elevator's official-type meter on split samples of corn and soybeans, at grain temperatures from near freezing to warm.

Why it is still open

Published tests look at one meter at a time. Acun and Gürsoy (2025) found that one capacitive meter's error on corn changed with moisture, temperature and variety. A preprint posted in October 2026 read one low-cost capacitance meter on 82 maize samples from 28 markets in Benin and found a root mean square error of 3.63 percentage points against the oven, where the regulatory tolerance is 0.8. NIST Handbook 159 gives weights and measures officials oven-tested reference samples for checking the commercial meters buyers use, not farmers' handheld ones. A search of Crossref and Europe PMC found no comparison across the common handheld models.

semanticscholar.orgapi.crossref.orgnist.gov

Built with what exists

Five handheld meters bought retail; a laboratory oven and a 0.001 g balance for the air-oven method; official moisture results on submitted samples as a second reference; a freezer and a warm room to set grain temperature; Python with pandas.

law.cornell.edu

  1. First step

    Buy the two most sold handheld meters, split 20 corn samples, and send half of each for official moisture as submitted samples.

  2. Done looks like

    Five models read on 150 split samples between 13% and 30% moisture at about 2, 15 and 30 °C; each model's average error and the range holding 95% of its errors against the oven and against the official meter, by temperature; raw readings public.