GeoAI-Driven Crop Yield Forecasting
Try now and be the first to try our product!
1. Mobile Version: GeoAI-Driven Crop Yield mobile App
To download our mobile app, click the icon below or search for “Super GeoAI” on Google Play or the Apple App Store.
Notes:
- The crop yield mobile app currently supports wheat, with additional crops like canola and peas coming soon.
- The iOS version is available now on the App Store. The Android version can currently be downloaded exclusively from our website and may be available on the Google Play Store soon.
Workflow and User Guide
- Preparation: Please ensure you have a frame such as 3×3 foot, a white background plate or white paper, and a device to separate the wheat head for photographing.
- App Setup: Ensure you have internet access to download the app and complete the registration and login process. Internet access is only needed for these initial steps. Once the app is installed, the mobile app and AI algorithms will function without requiring an internet connection.
- Site Selection: Randomly place the 3×3 foot frame on the field.
- Field and Site Setup: Use the “Add Field” and “Add Site” icons to set up the field and site where you’ll be taking photos.
- Counting Total Wheat Heads per 9 sqft: Navigate to the “Heads” tab to take photos of the wheat heads. For optimal accuracy, we recommend taking at least six photos within the frame, ensuring all wheat heads are captured without omission. Our algorithms will automatically aggregate the wheat heads across multiple photos within the frame. Each head should be photographed only once, and any heads outside the frame should be excluded. In future versions, our AI will automatically disregard wheat heads outside the frame.
- Counting Average Spikelets per Head: Click on the “Spikelets” tab to photograph the wheat heads and calculate the number of spikelets per head. For greater precision, each photo should focus on a single wheat head. To ensure optimal spikelet counting, please capture photos from a lateral view. When photos are taken from a front view, the AI will automatically double the count, assuming the opposite side has an equal number of spikelets. You can click on the spikelet result in the photo to switch between lateral and front view. If multiple photos are taken, our AI will select one to count. This step aims to determine the average number of spikelets per head for the site.
- Counting Average Seeds per Spikelet: Select one wheat head and count the spikelets. Then, remove the seeds from the spikelets and place them on a white plate or paper. Click on the “Seeds” tab and photograph the seeds. Enter the number of spikelets for that wheat head. This step will help calculate the average number of seeds per spikelet.
- Calculating Average Yield per Site: Click on the “Yield” tab. The algorithm will automatically display the total number of wheat heads, the average number of seeds per head (seeds per head = spikelets per head * seeds per spikelet), and the average crop yield for the site. Adjusters can manually modify these numbers if needed.
- Viewing Crop Yield Results: The average crop yield for the entire field, calculated across all sites, will be displayed on the dashboard.
Limitations:
- This version of the mobile app utilizes simple yet effective local AI algorithms. While the app delivers solid performance, integrating advanced AI processing on our servers could boost accuracy by an additional 5-10%. Currently, this version does not support photo uploads to our servers, but this feature will be available in future updates.
- The current AI algorithm has been trained exclusively on data from a single wheat field in Saskatoon. We have not yet incorporated training data from Alberta fields. Once local wheat data from Alberta is included in the training process, accuracy will further improve.
- For optimal spikelet counting, please capture photos from a lateral view.
- For accurate seed counting, ensure the seeds are clean and placed on a white surface, such as paper or a plate.
Crop Yield Formula
Method 1: To calculate crop yield, use the following parameters: row spacing (inches), heads per foot (ft), kernels per head, TKW (g), and harvest loss percentage.
Note:
- The conversion factor 1.15242 assumes that a bushel of wheat at 13.5% moisture weighs 60 lbs, and 1 acre equals 43,560 ft². This formula allows users to enter the weight (lb) per bushel for different types of crops. The default is 60 lbs per bushel for wheat and peas, and 50 lbs per bushel for canola.
- Typically, the average number of wheat kernels per wheat head is 28, and the Thousand Kernel Weight (TKW) is 28 g. This formula allows users to enter new values as needed.
- While the formula is originally for calculating wheat crop yield, it can also be applied to other crop types. For crops such as canola and peas, use “pods” instead of “heads” and “seeds” instead of “kernels.”
Although various formulas may be used for different crops, the above formula is one of the most accurate and straightforward methods for calculating crop yield in a field. This formula is adapted from the following reference:
- South Dakota State University: https://extension.sdstate.edu/sites/default/files/2020-03/S-0005-29-Wheat.pdf
- You can download the reference paper HERE.
Method 2: count by sqft
2. Drone version: Crop Yield Map
Crop Yield Zone Map
With our GeoAI-driven drone flight control application, a drone can autonomously survey a 160-acre field section within 15 minutes. Upon uploading the drone image folder to our GeoAI-driven cloud platform, the platform automatically stitches the drone images into a large, comprehensive image of the field. Our AI-driven system then generates a crop yield zone map in bushels per acre, identifying areas with high, medium, and low yields, as well as bare and weed-infested areas.
High-Resolution Crop Yield Contour Map
Click the link below to view the sample image and corresponding AI results.
This video demonstrates how SGA’s AI creates high-resolution wheat yield contour maps.
3. Web Version: Upload Smartphone or Drone Image to Calculate Crop Yield
Web URL: https://yield-ca.sga.ai/
This video demonstrates uploading wheat smartphone and drone images to automatically calculate the wheat crop yield in the images.