green-on-brown

Green-on-Brown Technology for Precision Weed Control

Green-on-Brown Technology for Precision Weed Control

While we have been working with green-on-green technology for a while and with great results, Proofminder prides itself on offering solutions for any and all Agroindustry challenges. Green-on-brown spraying is no exception 

The green-on-brown system can identify weeds on fallow fields and address them through spot spraying, relying on color differences to distinguish between weed-infested and clear zones.  

The advantages of green-on-brown spraying: 

  • This spraying technology is an environmentally friendly solution for farmers as it meets European regulations and follows global sustainability trends.  
  • Chemical usage can be decreased by up to 70% through spot spraying.  
  • By reducing the amount spent on herbicides, chemical plant protection can be profitable in crops whereas previously that may not have been the case.  

There are several possibilities in terms of use: 

  • Spraying stubble. 
  • Spraying corn or sunflower line spacing. – The challenge is distinguishing between the cultivated plants and the spacing.  
weeds on the field after harvesting

Figure 1 – Weeds on the field after harvesting

prescription green-on-brown weed map for spot spraying

Figure 2 – Prescription map for spot spraying after harvesting

How will green-on-brown spraying be profitable? 

 The below examples are estimates for a 20-hectare field:  

  • Total herbicide use costs 11 EUR / liter – the recommended dose is 5 liters/hectare. 
  • The herbicide cost for the entire 20-hectare area would come to a total of 1100 EUR. 
  • If, for example, only 40% of the field is weed-infested, then – per scale – the cost would only be 440 EUR.  
  • In this example, the agronomist can save 660 EUR on herbicides.  

 

geen-on-brown weeds on field

Figure 3 – Weeds on the field 

geen-on-brown weed map for VRA spraying

Figure 4 – Prescription map for variable rate application (VRA) of inputs

If you, too, struggle with weeds on your fallow fields, reach out to us for a cost-effective and sustainable solution.  

 

Let's talk about your project

Fill in the form and we will contact you shortly

green-on-green spraying technology

New technology on the horizon – “green on green”

New technology on the horizon - "green-on-green"

Precision farming is continually evolving, with ongoing advancements in AI computer vision, specifically within the field of “green on greentechnology. In previous years, the “green on brown” technology was widespread. This system could identify weeds on the fallow fields and address them through spot spraying, relying on color differences to distinguish between weedy and clean zones. 

In contrast, “green on green” technology differentiates between weeds and cultivated plants based on leaf morphological characteristics and other visual factors. This system can detect weeds in green crops. 

In practice, this involves creating a weed map athrough AI analysis after the drone survey.  This map shows the weed species composition in the  area, providing farmers with precise information on the location of weeds. Armed with this data, farmers can make well-founded decisions on mixing chemicals, areas to treat and the pace of application depending on the infestation levels of different zones.  

In a recent survey of a pea (Pisum sativum) field, several  weed categories were identified: ragweed (Ambrosia artemiisifolia), common milkweed (Asclepias syriaca), jimsonweed (Datura stramonium), german chamomile (Matricaria recutita), amaranth family (Amaranthaceae sp.). 

One of our clients, a major player on the food market, approached Proofminder to help them  fight a significant jimsonweed issue. High resolution drone images were collected by our partner, a professional drone service provider in Hungary Drone Deer, uploaded to Proofminder’s platform and analyzed by Proofminder’s AI model Proofminder not only detected jimsonweed and provided the GPS coordinates of the weeds for precision spraying, but also identified and located the four other most common  weed species. This information enabled the client to plan accurate  weed management actions, only applying chemicals in the areas where it was necessary. Due to these actions, the client was able to significantly improve food safety  and reduce the cost of spraying  dramatically. Identifying different weed species and distinguishing between similar weeds in a dense and lush vegetation was previously challenging, however, Proofminder’s platform now provides never-before-seen accuracy in weed maps that can help you in making the best weed management decisions.  

green-on-green technology for weed detection

Figure 1 – Jimsonweed and other weeds identified in green pea by Proofminder AI model

Furthermore, Figures 2 and 3 depict the AI-recognized location and density of the weeds 

green-on-green technology

Figure 2 – Various weeds and their GPS coordinates and density identified by AI 

green-on-green spraying technology

Figure 3 – Green-on-green technology for weed recognition and further hyper-precise spraying

Proofminder strives to provide innovative, cost-effective, environmentally friendly, and time-saving solutions to farmers. 

How is this achieved? 

  • Chemical usage can be decreased by up to 70% through spot spraying. 
  • The cost of chemicals can be decreased. 
  • Resistant or harder-to-treat weeds can be identified, allowing for the targeted use of larger quantities of herbicides or more expensive chemicals. 
  • Machine operating costs can be optimized. 
  • Reduced fuel consumption leads to lower carbon dioxide emissions. 
  • Time and resources are saved by avoiding unnecessary spraying.  
  • Weed maps and spray logs allow year-to-year comparison. 
  • Tracking the presence of weeds during different spraying cycles becomes possible. We can tell which weeds are more difficult to treat or even resistant, and if any have freshly appeared.  
  • Based on the above, spraying times and rates can be optimized in different areas.  

 

Contact us today to try out the new AI model for green-on-green identification and improve your production processes. 

Let's talk about your project

Fill in the form and we will contact you shortly

Weed map for hyper-precise spot spraying

Variable-rate spot spraying with AI-generated map and traditional sprayer

Cost saving - variable-rate spot spraying with a traditional sprayer? Yes, it is possible with the help of Proofminder!

Precision spraying is not only possible with drones and high-tech self-propelled sprayers. It is achievable for everyone with minimal preparation and a small workaround. If you have GPS and a traditional sprayer which is capable of section control, we have good news for you: you have the equipment to do spot praying and save up to 70% on chemicals.  

Proofminder offers a solution for growers with traditional sprayers that will help them engage the full potential of their equipment, save significantly on weedkillers, and at the same time, reduce the impact of chemicals on the environment. From agronomical and economical perspectives, every drop of chemical that doesn’t land on the targeted weed is a waste. As such, it would be counterproductive and wasteful to use the same rate of chemicals on the entire field. The rate should be proportionate with the density of the targeted weeds. If we can reduce chemical usage by 50% on every hectare, we can spray twice as large of an area with the same amount and since fewer refills are needed, we can also save precious time. This also means cost saving in economically challenging times 

Creeping thistle detection in corn

Figure 1 – Creeping thistle (Cirsium arvense) identified by AI in corn

If you own a commercially available drone, or you have access to a drone service provider in your area, all you will need to do is capture your field (we will provide the necessary specifications for the flight) before spraying and upload the pictures to Proofminder’s platform. Proofminder has a unique cloud-based platform which uses AI technology to identify different type of weeds and extract their coordinates. Using this tool Proofminder can generate a weed density map with the exact GPS coordinates of the different weeds. You may also filter for several types of weeds to decide which chemicals should be applied on different parts of your field. 

Creeping thistle identified by AI

Figure 2 – Creeping thistle spots identified by Proofminder AI model

As a next step, Proofminder exports the weeds’ coordinates to a format which is compatible with your traditional sprayer. You are now ready to utilize our spraying map with your machine. Enjoy your savings of time and money.  

John Deere spraying

Figure 3 – John Deere sprayer. Source: www.deere.com

We performed a field trial in Hungary, focused on spot spraying. A drone service provider captured images of a larger farm and uploaded the images to Proofminder’s platform. The automated evaluation and result generation took place during the night after the image capturing mission, and the customer received the results the next morning.  

The team responsible for spraying could now load the spraying map into their tractor which is operating a Horsch sprayer and is equipped with the John Deere controller.

Thanks to the spraying map provided by Proofminder, only the weed-infested portion of the field (67% percent) was sprayed and with that, 480 EUR/hectare was saved on chemicals and man hours.

This solution can be applied to several controllers of different manufacturers, and Proofminder is constantly working on extending the list of compatible controllers.  

Pictured below, are two neighbouring fields with different weed densities where the mentioned field trial took place. The amount of chemicals applied is directly proportionate to weed density. The field on the left was only slightly infected with creeping thistle and as such, only a smaller portion of it had to be sprayed. However, the field on the right was heavily covered with thistle, but the weed density was varied, therefore a custom rate of chemicals was applied to save on chemicals, which could not have been achieved if the entire area would have been sprayed with an even amount.  

Prescription map for variable spot spraying

Figure 4 – Prescription map generated by Proofminder AI model for variable spot spraying with traditional sprayer

Contact us today to implement the precision spot spraying and save your budget. 

Let's talk about your project

Fill in the form and we will contact you shortly

Spot spraying using DJI Agras and Proofminder weed map

Johnson’s grass and other weeds in corn: how to save up to 70% on chemicals

Johnson grass and other weeds in corn: how to save up to 70% on chemicals

Weed control plays a vital role in plant protection. Weeds represent a threat to cultivated plants, as they might shade them, drain nutrients and water from them and even hinder their ability to germinate.  Furthermore, weeds can cause health issues and may also impede the harvesting process. “About 100s of euros of yield loss per hectare caused by johnson grass” – a production challenge that the grower raised and invited Proofminder to help.

Proofminder’s trained AI model can identify numerous weed species, such as johnson grass (Sorghum halepense), creeping thistle (Cirsium arvense), and jimsonweed (Datura stramonium) for example. In this case study, we will be focusing on the johnson grass recognition AI model in hybrid corn. 

Corn is extremely sensitive to early weeding, as its ability to suppress weeds is still weak at that stage. Controlling johnson grass – one of the biggest threats to corn – is very important, as, if left uncontrolled, it can result in significant yield losses due to its ability to drain water and nutrients from the cultivated plants. Furthermore, johnson grass can be a host for the Maize chlorotic dwarf virus and Maize dwarf mosaic virus, and its allelopathic effect is also significant. 

This case study focused on a 26.8-hectare hybrid corn field plagued by a johnson grass infestation. During the application of the traditional field spraying technology, the entire field is sprayed with a specified dose, averaging 200-300 litres per hectare. However, this approach is less aligned with the latest precision farming trends. Spot spraying can be a much more environmentally friendly and cost-effective solution, since many weeds, such as creeping thistle, occur in localized patches. Targeted spraying of these weed-infested patches is more easily applied and efficient as using pesticides should be sufficient.  

  • A standard surveying drone (either the grower’s drone or Proofminder drone operator partner) conducts a comprehensive survey of the field, capturing high-resolution images.
Johnson weed in corn

Figure 1 – Johnson grass in corn, image from DJI drone

Johnson grass in corn

Figure 2 – Johnson grass in corn, field overview 

  • With the help of our trained AI model, the captured images are analysed and the weeds are distinguished, with high precision, from cultivated plants. This analysis also shows the extent of the infestation on the field.
Johnson grass and corn plants identified

Figure 3 – Proofminder AI Model for johnson grass recognition

AI model for Johnson grass recognition in corn

Figure 4 – Proofminder AI Model for johnson grass recognition, field overview

  • Based on the analysed data, a usage map is created, and presented in a standard geoinformatics format, such as a shapefile, which can then easily be downloaded to a machine sprayer or spraying drone.  
Weed map for hyper-precise spot spraying

Figure 5 – Proofminder weed map for hyper-precise spot spraying

Spot spraying using DJI Agras and Proofminder weed map

Figure 6 – Spot spraying with DJI Agras T30 using Proofminder weed map

  • Out of the total area of 26.8 hectares, only 7.3 hectares 74 precise spots) had to be sprayed with graminicides. Spot spraying was made possible with a drone in this case. As weather conditions during the day were windy, which is less than ideal, spraying took place at night. The spraying mission was a success as no spray drift occurred, and 112 EUR/ha worth of chemicals could be saved. 

Reach out to us to control weeds, save 40-70% on fertilisers, and stay ahead of the curve. 

Let's talk about your project

Fill in the form and we will contact you shortly

Detasseling in hybrid corn

Detasseling: from manual labor, days on end – to drones and AI

Detasseling: from manual labor, days on end - to drones and AI

Scientists and growers started experimenting with hybridizing seed corn in the early 20th century, and even back then, detasseling played a most significant role in the production process. In a mere century, we went from extensive manual labor, to introducing machined detasseling, to eventually having access to the latest technologies of drones and artificial intelligence.

Why is detasseling crucial in hybrid seed corn production?

Precise and timely detasseling – the removal of the tassels from all corn plants of one variety – results in a more physically uniform batch, increased yield, and higher genetic purity. Achieving a marketable purity rate of 99.7% leaves minimal room for error. The detasseling process is highly labor-intensive and very costly. This is where Proofminder steps in.

Hybrid seed corn field after detasseling

Figure 1 – Hybrid corn seed field after machine detasseling

Missed tassel in hybrid corn

Figure 2 – Missed tassel

Challenges addressed by Proofminder

  • Proofminder accurately identifies missed tassels and provides their exact GPS coordinates. We provide a kml file (Google Earth file) that can easily be viewed on any device. These files can provide helpful insights for agronomists, seed production managers, as well as manual detasseling crews to prioritize areas with missed tassels.  
  • Proofminder helps reduce labor costs and time requirements.  
  • Our solutions save cost and increase the yield.  
  • We offer an environmentally friendly and sustainable approach to detasseling projects.  
  • Our solutions can be seamlessly scaled to thousands of hectares of land.  
Missed tassel detection in hybrid corn

Figure 3 – Missed tassels identified by Proofminder AI Model and marked on the field

Missed tassel map in hybrid corn

Figure 4 – Missed tassels report built with Proofminder

Our process 

  • A customized drone flight is planned specifically for your field. The drone’s camera is angled so that a much larger area is visible on the captured images.  
  • Commercially available, off-the-shelf drones are adequate for tassel detection missions.  
  • Proofminder can also arrange the drone flight part for you through our global network of drone service providers.  
  • A single drone can capture and process over 100 hectares in a day.  
  • Captured images are uploaded to the Proofminder platform.  
  • Our trained AI model identifies male and female rows, enabling the exclusion of male plants during the missed tassel identification phase.  
  • We provide comprehensive reports, including exact GPS locations and visualizations of missed female tassels. The GPS coordinates in the provided shapefiles can be loaded onto handheld GPS devices.  
  • We deliver results in less than 24 hours. Since time is of the essence when it comes to detasseling, we capture images during the day and provide actionable reports before 6 AM next morning. This way, you can plan the day of your detasseling team in the most efficient way.  

Figure 5 – Proofminder report: field overview with exact GPS coordinates of each missed tassel

Coordinates of missed tassels of hybrid corn

Figure 6 – Exportable reports for the detasseling team

Your involvement  

  • Image Capture: If you opt to take the images yourself, we will require high-definition images taken at a specific height and angle. We will provide the necessary specifications and handle the rest once we receive the images.  
  • Proofminder’s Image Gathering Service: Alternatively, if you prefer, we can manage the entire image gathering process. All we need is a contour of the field in question and information regarding the sowing direction.  
Image gathering with DJI drone

Figure 7 – Image gathering, flying over the corn field 

Drone image gathering for missed tassel identification in corn

Figure 8 – Image gathering using DJI Drone

About Proofminder 

Proofminder enables Agroindustry players to transition to plant- and leaf-level farming. Leveraging AI to extract actionable insights from drone images, we provide growers with invaluable information about every square centimeter of their fields, across the entire season.  

We collaborate with both growers, including agriholdings and farmers, cultivating corn seeds, as well as seed producers. Numerous top seed producers already entrust us with their detasseling efforts.  

Don’t hesitate to reach out to us to address your detasseling challenges and stay ahead of the curve.

Let's talk about your project

Fill in the form and we will contact you shortly

pine tree counting with a drone and AI

Beyond the visible: Gaining Vital Insights for Pine Tree Nurseries with DJI Mavic M3 Multispectral and Proofminder AI.

Beyond the visible: Getting crucial insights for pine tree nursery with DJI Mavic M3 Multispectral and Proofminder AI model

When it comes to running a tree nursery, there are a lot of factors to consider in order to maximize efficiency and yield. 

Direct field observations and timely insights on tree nurseries will always be essential for tree management to:

  • Get reliable information on the exact number of trees
  • Ensure that the nursery is able to meet demands
  • Reveal and track germination problems
  • Detect weeds and plant health problems before they manifest over larger areas
  • Visually capture trial progress and outcomes
  • Forecast yield and income at various stages
  • Estimate the irrigation effectiveness, damage assessment and more.

Challenges and current methods of tree calculation in nurseries  

Right now, tree nurseries rely on a time-consuming method of manual field measurements. This means physically sampling each species, by stock type and seedlot in the field, estimating tree size distributions and, therefore, their value.  

The challenge is, there could be multiple stock types across a dozen species with multiple seedlots per species, each with different parameters and pace of growth. Even for a small nursery operation of 45 acres (~18 hectares), the manual process of examining plants could easily take several days in the field not to mention time spent analyzing the numbers and figuring out if there will be shortfalls or excess plants to sell. With so many variables and demand for trees only set to increase (Fargione etal, 2020) new approaches are needed to address inventory and production challenges. Expanding nursery operations is no simple task, making nursery operations more efficient and productive is within our reach. 

A Game-changing approach: utilizing DJI Mavic 3 Multispectral (M3M) and Proofminder leaf-level farming platform to maximize the efficiency of tree nursery management.  

Using drones and specialized software allows the capture of high-resolution images and data not only to calculate the number of trees but also their health and size distribution. This method is faster and more accurate, in the short term allowing nurseries to better manage their inventory to make informed decisions about sales, pricing, and order fulfillment. In the long term, these new approaches could help nurseries gain insight into field soil health, irrigation and seedlot performance, and early detection of plant health issues for effective diagnosis and intervention. 

The use case described below is built for a tree nursery in the Pacific Northwest of North America, in collaboration with nursery and forestry professionals.  

The germination rate, plant health, and size of the trees (principally height and crown cover) are core elements of optimized yield in a tree nursery.  At the use case site, the nursery operates on 45 acres (~ 18 hectares) annually. During the 2023 growing season, Proofminder and its collaborators are evaluating images of 14 conifer species, across 2 stock types, and representing over 35 seedlots. Nursery professionals do their best to assess the stem caliper, stock height, and individual tree counts to ensure they meet customer specifications and maximize the overall yield. Accelerating data gathering and analysis could make a real difference to the tree nursery business from seed germination to seedling sales. 

The first stage when the tree nursery begins manual observations is when the plants have just emerged.  

tree counting in the pine tree nursery

Figure 1 – (left image) emergent field sown Douglas-fir (Pseudotsuga menziesii) up close and (right) planting bed rows after the emergence are captured on a phone camera. 

Trees at the emergent stage are approximately 1-2 inches (~3-5 cm) tall, too small to detect with standard visual imagery but the near-infrared and red-edge bands of the DJI M3M are able to detect even these small emergent seedlings. 

The image below taken with DJI M3M reveals the areas of emerging trees and germination from the trees sown from seed. These early images are an indication of how well the sowing technique worked and how well the emerging plants are growing relatively to current irrigation. 

Imagery for tree nursery taken with DJI M3M

Figure 2 – Imagery for tree nursery taken with DJI M3M. 

In addition to routine field observations to identify irrigation issues the tree nursery periodically and systematically monitors plant growth and health throughout the growing season to identify problem areas with weeds or damage caused by insects and pathogens. Beyond the growing season, the nursery is monitoring the weather to determine when to begin the annual harvest operations, referred to as ‘lifting season’, during this time the nursery may also need to assess crop losses due to the effects of severe frost damage 

pine tree counting with a drone and AI

Figure 3 – Image (left) Ponderosa pine (Pinus Ponderosa) trees taken from eye-level with a camera phone and the image (right) captured the same day by the drone with individual trees detected by Proofminder AI Model. The pine trees were sown the previous year and are entering their final growing season at the nursery before being lifted and packaged for delivery to customers around the Pacific Northwest.

High quality image capture is necessary for individual tree detection. The M3M is equipped with real-time kinematic GPS. However, precision GPS needs to be paired with appropriate flight parameters. In tree nurseries, there is a need for experimentation on timing image acquisition given the effects of crown closure on individual tree detection. More testing is anticipated with oblique imagery and video to further enhance image analysis outputs and reduce image processing of each individual multispectral band. Other parameters such as flight height are influenced by the need to get sufficient resolution (i.e. ground sampling distance) to distinguish the individual tree tops, which can vary due to the crown and leaf morphology of a given species. Flight speed and image overlap needs to balance efficient image capture with consistent ambient light to ensure dense reconstruction during the image processing. In this case, with several smaller fields (~1-2 ha) that contain conifers flight heights can range from 40-60 feet (~12-18m) with 80% overlap produce consistent results.

Figure 4 – Video of image gathering process to analyze early growth stage of Oregon White Oak (Quarcus garryana) with DJI M3M. 

Additional insights can be revealed with the Proofminder leaf-level farming platform and its AI models. The report below generated with an AI model for precise tree stand counting run with multispectral images. Every tree is marked on the field and has its own GPS coordinates. Also, the model could distinguish plants by phenotype, identify weeds, missed plants, distance between plants and deliver actionable information to the grower.  

Plant counting report generated by Proofminder AI model for tree plant recognition

Figure 5 – Plant counting report generated by Proofminder AI model for tree plant recognition 

The quick summary and results of using DJI M3M and AI 

  1. Time savings. As mentioned above, the drone flight time could be about 2 hours plus a few minutes to process images through the Proofminder platform to get lots of valuable reports vs. 2-3 days of manual sampling. 
  2. Convenience. It simply might be impossible to physically go to the field after the rain, but using a drone allows one to overview the whole field remotely or estimate losses. 
  3. Accurate data. The high-quality multispectral images plus trained AI models deliver actionable and timely reports. Detailed examination of trees manually usually contains mistakes, is highly dependent on the knowledge of the employee, and is barely scalable for larger areas. The accuracy of data and speed of report delivery are also crucial factors for the decision-making process. 
  4. Increased efficiency and yield. As a result of all the above benefits and proper reports, nurseries can quickly realize their inventory status, make adjustments in sales and orders based on reliable data. As an example, the height of the trees in a nursery is one of the crucial factors that directly affects its income. There are 3 sizing “buckets”: High, Medium, Low. All the ‘Low’ trees i.e., below 6 inches (~15cm) simply will not meet most customer specifications. In many cases, the most desirable trees heights are between 10-18in (~25-45cm) closest to the medium size range since the shoot-to-root ratio will likely be more balanced and thus better suited for outplanting on a wider range of sites. 
  5. The details invisible to the naked eye. Multispectral imagery gives agriculture professionals a way to see the invisible details such as plant health indicators that cannot be seen by visual inspection. The proper interpretation of data, accurate reports and timely actions of growers can prevent yield losses and help to grow more with confidence.  

Contact us today to try out the AI model for tree counting and get actionable insights for your tree nursery or orchard.

Let's talk about your project

Fill in the form and we will contact you shortly

Precise plant stand count with drone

High-precision plant stand count for corn, sunflower and sugar beet with a drone and AI

High-precise plant stand count for corn, sunflower and sugar beet by a drone and AI

This article covers precise plant stand count using an off-the-shelf drone and Proofminder’s trained AI algorithm for accurate yield assessment and the following insights on the field. You will find practical tips on image collection and recommended approach for corn, sugar beet and sunflower, but the information is also useful for other field crops, vegetables and orchards. Keep reading!

Plant stand count is an essential task in yield management. It allows growers to estimate the plant population, density, germination rate, and plant health and make timely decisions that finally affect the yield. Common manual methods of plant stand counting have helped growers for decades. They are based on visual inspection and plant calculation on small pre-defined field areas. However, these methods are laborious and far from accurate. Fragmented plant stand count does not provide the complete picture, and problem areas with uneven emergence or weeds might be overlooked. The lack of information on the field eventually leads to a waste of resources and less profitable decisions.

New technologies like drones and AI leverage the opportunity to make Agri operations smarter and more efficient. With this innovative approach, growers can now receive accurate data, make timely decisions and sustainably maximise the yield. Surprisingly, this is not as complicated or costly as it might seem.

This article covers precise plant stand count using an off-the-shelf drone and Proofminder’s trained AI algorithm for accurate yield assessment and the following insights on the field. You will find practical tips on image collection and recommended approach for corn, sugar beet and sunflower, but the information is also useful for other field crops, vegetables and orchards. If you have a drone or considering buying one to turn a tedious task into an interactive process and get a high-precision result, keep reading. You will find drone requirements, flight tips and common mistakes, and learn how to get a precision stand count report in a few hours with an innovative AI farming platform.

Why and when do you need a precise plant stand count?

There are situations when a low accuracy report is acceptable, but it is absolutely essential to have a precise one if you aim to:

  • Check the sowing quality, especially if you are producing seeds;
  • Understand zones of varying productivity in the fields;
  • Receive accurate data during R&D projects;
  • Estimate the yield precisely in the early stages;
  • Spot rogues;
  • Make timely decisions, i.e., partially replant the field;
  • Increase the yield potential to meet the production goals.

When is the best time for plant stand count using a drone and AI?

Estimating the number of plants and their density is crucial for early-season yield management. The accurate information here is a chance to save the yield if something goes wrong and improve the harvest. To gather proper images for further analysis, consider the tips about plants and the weather. 

The plant should be big enough to be seen from the air, but the leaves are not yet too close to each other to distinguish plants and estimate the density. As an example, for the precise stand count of corn, the plant should have about 3-7 leaves (V3-V7 vegetation stages). The weather should be stable during the footage, thus the lens can adapt to the conditions whether it is sunny or cloudy. Also, it should not be too windy, note that the wind speed may greatly vary depending on the altitude. Which altitude is right for a stand count? Find below!

Manual plant stand count of corn

Figure 1 – Corn field 

Common method of plant stand count of corn

Figure 2 – Manual plant stand count of corn

Capturing images by a drone – instructions and tips

The ideal resolution for plant stand count by a drone and intelligent software depends on the plant and the goal. For precise stand calculation of corn, sunflower, sugar beet, and some other field crops and vegetables would be 0.8 cm per pixel or less. What does it imply, and what kind of drone is suitable? The widely available DJI Phantom 4 Pro V.2. can be a good entry-level option for that job, similarly, the DJI Phantom 4 RTK is also a great option if you want a professional drone with high precision positioning. You will need to fly at 18-30 meter altitude to get the indicated resolution. Be aware that some of the Integrated controllers (the Plus versions) limit the flight altitude to 25m above the ground so if you want to count small crops and fly low, you would rather choose the simple controller and instruct the drone from your mobile or tablet.  The ideal speed to capture detailed images would be between 3-5 m/s depending on the altitude and the wind conditions. Using this drone, you can proceed at about 25-30 hectares per day if you have enough batteries; mind you: you can charge them on the site.  At Proofminder, we are working on novel ways to do this image capturing and foresee the possibility in the near future to capture up to double of this area per day by a Phantom 4 drone.

Figure 3 – Shooting images for plant stand count by DJI Phantom 4

Things to avoid; the Top-10 common mistakes in drone footage:

  1. Wrong exposure setting, not properly assessing the weather, resulting in over- or underexposure. Overexposure is more of a problem than underexposure, so if you need to choose between cloud and sunny, and you are not sure, you can safely go for sunny.
  2. Too much wind or unstable weather conditions result in blurry images.
  3. Not equipped with sufficient memory cards, make sure you have at least a 64 GB card for ~40-50 hectares of land.
  4. Not enough batteries and/or chargers to fly continuously during the day.
  5. Shooting after rain may require some recalibrations because the plant on the wet soil may not be visible enough, keep this in mind.
  6. Not flying with the right amount of front/side overlaps, potentially preventing stitching pictures together and creating an orthomosaic. 75% is a safe value in most cases.
  7. Flying too fast results in blurry images.
  8. No right logistics and setup – i.e., make sure you have a suitable car and path to access the field, have a generator available to produce power for all the equipment, battery charger and laptop, have a sun-shaded place to work from, etc.
  9. No proper preparations in flight planning – e.g., cater for height differences in the field upfront.
  10. Check the airspace before flying and make sure not to fly beyond visibility to avoid losing your drone.
The process of drone footage for precise stand count

Figure 4 – The process of drone footage for precise stand count

The shape-file of the field built on Proofminder Platform

Figure 5 – The shape-file of the field

Plan stand count report and additional insights on your field

Following the instructions will result in lots of useful data and good images for further analysis and insights about the field and plants. What can you, as a grower, do with the collected images? There are a couple of ways – as an illustration, to analyse it manually, which is again time-consuming and subjective or use Artificial Intelligence, which can do the job quickly and accurately. The AI-powered platform can create an orthomosaic, an automatic plant stand count report and mark issues on the field that are not visible or not humanly possible to discover in traditional methods.  

Images below show what your plant stand count can look like on the Proofminder platform.

On the automatic report generated in a system, you can see

  • Plant & row density;
  • Precise plant stand count;
  • Each plant is marked on the field with precise coordinates;
  • Plant distinguished by phenotype, in this case – male and female plants of hybrid corn are marked with a different colour;
  • Zoom-in feature to analyse specific zones, rows or plants.
Precision plant stand count of corn

Figure 6 – A plant stand count report on the Proofminder platform

Plant-level stand count report on the Proofminder platform

Figure 7 – A plant-level view of a stand count report on the Proofminder platform

Additional insights & platform capabilities

  • During the corn plant stand count, we discovered that lots of plants on a field were destroyed by wild boars;
  • The problem areas can be marked with GPS;
  • Downloadable shapefile for further usage e.g., compare it with sowing facts;
  • As each plant has precise coordinates, derived metrics such as the distance of plants, density, gaps, row distance, etc. can be provided additionally;
  • Actionable insights on a level of leaf or plant.

Automated plant stand count - outcomes and benefits

  1. The plant stand count accuracy of manual methods is hard to estimate; one thing is clear: it can only be precise on small analysed areas of the field; applying these numbers to the whole plot would not give precise information. Drones and AI technologies are able to provide growers with 90-99% of stand count accuracy and reveal other problems on the plant level.
  2. Technologies make the plant stand count process way more precise, interactive and insightful.
  3. Additional insights discovered: lots of plants have been destroyed by wild boars.
  4. A helpful opportunity to export the stand count report and reuse its data in other farming activities.
  5. Possibility to get the maximum out of the drone-made images and use this information for data-driven decisions and for growing more with confidence.
  6.  

Let's talk about your project

Fill in the form and we will contact you shortly

Tomato ripe check

10+ efficient AI computer vision use cases for higher yield

10+ efficient AI computer vision use cases for higher yield

There is a great need to improve crop productivity, and AI is doing it quickly and efficiently, not just in laboratories but on actual Agri holdings!

We have collected great examples of how intelligent algorithms help crop production and increase growers’ productivity during the season, from pre-planting to post-harvest. All the use cases listed require visual data (preferably from drones) and trained AI models to solve particular issues. In a nutshell, visual data provided by drones are the “eyes” of the operation, and AI and machine learning intelligence is the “brain,” together, they are changing the Agroindustry making it more productive and efficient. 

This article covers AI application for crops and vegetables such as cereals, corn, soy, wheat, sugar beet, and more, but specific use cases for orchards, indoor farms, and livestock are also feasible and beneficial, this is just a topic for another discussion.    

All is visible on the field during the season can be measured, estimated, and improved with the help of AI.

Agriculture cycle

Agriculture cycle

Crop planning

1- Waterlogging and field relief analysis. 

These details most of the time cannot be precisely defined with the human eye or satellite images. Drone images and AI can provide farmers with insights that can bring value even before the first plants appear on the field. Knowing the field’s specificity will help support numerous upcoming decisions on seeding, irrigation, inputs, etc.   

Definition of production zones

Planting  

2 – Stand count and yield assessment.

Review the precise number of plants in a specific field and make decisive actions if it’s under the norm or expected results. Evaluate seed quality (germination rate), decide whether to replant or not, assess the yield, and mark the zones of potential yield losses.   

3 – Plant density.

Since both high and low crop densities reduce yield and total revenue, it is crucial to know the situation on the field as soon as possible. Growers know there is a population for each crop that maximizes crop yield. Start strong! 

Precise plant standcount

Crop monitoring 

4 – Heading date/flowering stage detection. 

Accurate chemical thinning of some crops and fruit trees requires estimating their blooming intensity and determining the blooming peak date. With drone data, AI quantifies the size and number of flowers for timing fertilizer input.  

5 – Weed detection. 

Left uncontrolled, weeds can result in 100% yield loss. With vision processing and machine learning technology, today’s AI-driven equipment can reduce 80-90% of herbicide use. AI can distinguish and classify weeds with a high level of accuracy to support wise decisions.   

Weed detection in sunflower

6 – Disease detection. 

Up to 50% of yield loss is caused by pests and diseases every year. AI technology helps diagnose plant diseases, classify them and significantly decrease the number of chemicals applied.   

7 – Insect damage.

It is humanly impossible to distinguish insect categories and the growth period of insects without the knowledge of entomology. AI computer vision makes it possible to recognize and classify insects. It can quickly locate the information of an insect positioned in a complex background, accurately distinguish insect species with high similarity, and effectively identify the different phenotypes of the same insect species in different growth periods. This information will help reduce the number of insecticides applied and fight pests intelligently.  

8 – Nitrogen analysis.

The deficiency of this macro-nutrient directly affects crop health and yield. However, when Nitrogen inputs to the soil system exceed crop needs, excessive amounts of nitrates enter ground or surface water, thus leading to significant environmental harm. With the multispectral drone camera and intelligent, trained algorithm, staying informed and supporting decision-making on fertilizer quantity and timing is possible.   

9 – Crop monitoring and Plant stress analysis. 

Tracking healthy leaf color, plant growth, and ever-changing environmental conditions during the season requires significant time and human efforts. AI computer vision offers an effective alternative for growth stage monitoring because of its low-cost (relative to person-hours invested in manual observations) and the requirement for minimal human intervention, as a result – less costly mistakes and better yield.

10 – Specific tasks and quality control. 

As an example – missed tassel recognition during detasseling in maize seed production. Every tassel not removed by a machine leads to unwanted pollination and decreased genetic purity of maize seeds. AI can recognize missed tassels from 3 cm size before they start pollinating and help to save the whole field. 

Harvest & Postharvest  

11 – Plant size for harvest readiness.   

In practice, the crop growth stage is still primarily monitored by-eye, which is laborious and time-consuming and subjective, and error-prone. The application of AI computer vision offers a precise, cheaper alternative to manual observations. It is now easier to plan the labor needs, predict the best harvest day, and even refine revenue projections.  

Tomato ripe check

12 – Remaining plant sizes and count.   

Make sure that you do not leave the revenue behind. This second check and intelligent model will help identify the remaining plant counts and size. Last but not least – there are insights about quantifying crop residue, which help you with your further soil management strategies (like cover cropping or no-till planting).  

Final word. 

AI can provide growers with instant insights from their fields help them to identify areas that need fertilization, irrigation, or pesticide treatment. The AI technology combined with timely actions allows growers to reduce the number of fertilizers, help increase food production while minimizing resources, increase yield and profit, and make growers’ lives easier in the current labor-shortage situation. With valuable information, growers can make better decisions, recalibrate their future planting strategy, and finally grow more with confidence. 

AI use cases in Agroindustry

Let's talk about your project

Fill in the form and we will contact you shortly

Detasseling map of hybrid corn field

Leveraging AI in Hybrid Seed Production

Case Study: Innovation in hybrid corn seed production

Innovation in hybrid corn seed production: Identifying missed tassels during detasseling to increase genetic purity, decrease cost and prevent revenue loss

Hybrid corn seed  production problems

  • High chances of losing a field due to missed tassels in female inbred rows
  • High cost of manual detasseling due to reruns
  • The procedure must be carried out in a very limited time with the high accuracy

CHALLENGE

Machine detasseling only results in around 85% pulling vs the needed 99.7% for hybrid corn seeds

Any tassel not removed during the detasseling phase of hybrid seed corn production can result in unwanted pollination, degrading the necessary genetic purity and quality of seeds thus impacting their commercial market value.

SOLUTION

  1. Drone images uploaded to Proofminder platform. We collected data from 5 distinct fields and multiple corn varieties
  2. Field visualization. Orthomosaic of the plot is automatically created in the system
  3. High precision stand count readily established
  4. Male and female lines are distinguished based on phenology. Sowing structure, lines, distances between lines and individual plants identified without reliance on sowing or other external data. Male/female distinction allows to ‘ignore’ male tassels during the missed tassel identification phase
  5. Clear report ready. Actionable insights in the palm of our hands
  6. Yield saved. With timely actions taken by the agronomist, the cost of detasseling can be decreased by up to 1/3

DETAILS

Data Collection Approach:

  • Drone camera angle set at 60 degrees not at nadir – much larger plant area visible, training and verification is easier
  • Our algorithm calculates tassel location with 5-10cm accuracy based on drone location
  • Off-the-shelf drones (DJI Phantom, Matrice 210 v2 + Zenmuse X5S, etc.) and flight planning software (DJI Ground Station Pro) used
  • Above 100 hectares per drone per day coverage and processing possible

Identification of missed tassels:

  • Tassels categorized and predicted per size (S-XL)
  • Male tassels excluded, leveraging row identification algorithm
  • Identification is infinitely scalable using cloud resources – processing of 1 hectare on 1 node under 30 minutes
  • Identified missed tassels visualized in the Proofminder application on desktop/tablet/mobile but can be exported into geojson or other common geospatial formats

Take control of the detasseling process and stay a reliable seed supplier

Avoid losing a field due to missed tassels

Get higher genetic purity of the hybrid corn seeds

Decrease cost of detasseling up to 1/3 with less labor hours

Let's talk about your project

Fill in the form and we will contact you shortly

Ragweed_detection_with_AI

Case Study: Ragweed Identification​

Case Study: Ragweed Identification​

Ragweed (Ambrosia) identification based on phenology, using high-resolution RGB drone aerial photos

CHALLENGE

Ragweed pollen is a common allergen causing yearly suffering for millions of people. A single ragweed plant can produce up to a billion pollen grains in a season, which are carried long distances by the wind. It is a major health concern in many parts of the world – also, various countries have laws imposing fines if too much ragweed is found on a property.

ragweed_detection
ambrozia_detection_computer_vision

SOLUTION

Our partner turned to Proofminder to be able to recognize and identify ragweed on ragweed on thousands of hectares of land daily, based on plant phenology. 

It only took a few weeks for visual AI development and deployment on the Proofminder platform to ensure scalability.

Ragweed detection and the calculation of infection metrics became accessible via a simple, browser-based map display, to take decisive actions.

DETAILS

  1. Orthomosaic is automatically created
  2. Multiple Machine Learning algorithms were evaluated by leveraging Proofminder’s MLOps quick iteration capabilities. SVM and Random Forest were deemed most applicable
  3. Infection metrics are accurately calculated according to local law
  4. Country-level deployment envisioned using Proofminder:
  • 1st pass using satellite photos to identify potentially infected areas
  • 2nd pass uses drone photos for detailed analysis and ragweed identification
  • 3rd pass can focus on elimination – e.g. spraying
ragweed_detection