AI systems are helping to perform crop yield estimation, resulting in overall improvement in field resource utilization. Accurate and error free yield estimations are now at the tip of your fingertips with Sowit’s AI driven crop yield estimation technology at place. This post covers the case study of this futuristic farming technology powered by computer vision.

This article was originally published by Ultralytics.

If you didn’t already know, you know now – Ultralytics HUB is where you bring your ideas to life using AI.

You don’t need any previous AI experience. You don’t need to know any code. You don’t need to be a programmer. And you definitely don’t need to be the next Elon Musk. 

All you need is an idea… 

Just like these guys, the founders of SOWIT, who used Ultralytics AI to create an application that’s transforming the future of farming.

Who are SOWIT?

SOWIT is a team made up of agronomists, developers, and engineers… and we may as well throw in innovators while we’re at it. They also have an unbelievable ambition that’s ready to blaze new trails in agriculture: to empower African farmers by providing them with technology that will revamp the way they do agriculture. 

What was the problem SOWIT wanted to solve?

Farming is difficult. And doing it with minimal equipment is even more so. In the region, farmers often must go through laborious, taxing, and cumbersome processes to complete daily tasks. 

Between the Marrakech and Imintanout regions in Morocco, a citrus farm that SOWIT has been working with had some equipment at its disposal but not enough to maximize its full farming potential. 

This made estimating crop yields a grueling task that drained everything from resources to time. Put it this way…

…Imagine counting every single fruit on trees across 180 hectares of farmland.

It’s not easy. But if you were a farmer, this is something you must do to put bread on the table. And for the farm itself, this task had its own drawbacks:

❌ Time-consuming

Manually counting fruit across farmland takes up a lot of time that could be spent elsewhere. 

❌ Demands a lot of human resources

Having an entire workforce of farmers counting fruit is expensive and burns through the capital. 

❌ Workers are prone to error

Humans are susceptible to biases that can skew estimations significantly. One test highlighted that in 23% of cases, two farmers who’ve counted the same fruits will have a total difference of more than 10%.

❌ Farms lose money from inaccurate estimations

Manual yield estimations aren’t accurate, with the relative error above 30%. This level of inaccuracy creates a ripple effect that impacts all areas of performance, including the planning, managing, and selling of their products. Small farms in particular are always on the back foot when selling to buyers with stronger negotiating power. 

So how did SOWIT use Ultralytics AI to empower smaller farms?

The biggest challenge was being able to give farmers access to affordable technology that could yield the greatest results. And so, of course, SOWIT turned to Ultralytics and the results were exceptional.

SOWIT used YOLOv5 (You Only Look Once), the AI architecture that’s the backbone of Ultralytics HUB, to train their model to identify fruits scattered across a tree. YOLOv5 object detection models are famous for being the fastest and most accurate in the world, providing users with real-time object detection.

It’s hyper-efficient. It’s powerful. And it’s easy to use, and you can too with Ultralytics HUB. Think of Ultralytics HUB as your bridge to the AI realm and your introduction to object detection. 

You don’t need to know a single line of code to use YOLOv5 and to have fun with machine learning models. Train any model you want, to do anything you desire from any idea that inspires you. 

You can create anything at Ultralytics HUB. 

Thanks to SOWIT, farmers have a new, easy, and fast way of estimating crop yields. Farmers simply have to download the app, take 2 photos of the tree, and the number of fruits will be calculated in real-time. 

Below are some images of YOLOv5 in action. How long would it take for you to count every one of those fruits?

Farmer quote, Youssef Sridi, Technical director of the Mfruits farm: Using SOWIT’s application helped us accurately estimate our yields, in a timely manner with great precision, while saving field resources

How the vision AI app works

Step 1: Log into the SOYFIELD app

Farmers first have to log into their personal accounts. 

Step 2: Enter plot details

Each farm is unique. Farmers have to identify their plot and draw its contours on the app. Afterward, they’ll have to input basic information about their farm, including crop type, number of trees, cultivation practices, etc. 

Step 3: Provide tree samples

Farmers are asked to sample a number of trees to meet the required margin of error. This is done by taking two photos of each tree (front and back).

Step 4: Let YOLOv5 do all of the hard labor

Now, this is when the magic happens. All images are sent to the server for YOLOv5 to examine and identify the number of different fruits hanging from each tree. 

So what does this mean for farmers?

✅ More accurate yield estimations

Less bias means more accuracy. Farmers can rely on an accurate and consistent sampling methodology, which in the long run leads to an increase in revenue and profits. 

✅ Strong bargaining power over buyers

Accurate and consistent yield estimations allow farmers to estimate their yields up to two months before harvest. This provides them with greater leverage over brokers who visit the orchards to buy the fruits.

✅ Mass capital savings

Farms can cut labor costs by not having to send as many workers to count the fruit. This additional capital can be invested elsewhere on the farm. 

✅ Increased productivity and revenue

YOLOv5 provides a more automated approach by eliminating the need to manually count every piece of fruit on the farmland. Farmers can boost productivity by completing crop yield estimations in a fraction of the time. More productivity means more revenue. 

✅ More time to be invested elsewhere

Want to have fun with AI Computer Vision?

All you need is an idea. With Ultralytics HUB, it’s easy to create models with YOLOv5 and bring your ideas to life. We make things simple and we do all of the heavy-lifting, so you don’t need to know any code to have fun AI. It’s easy to get started and even easier to build your model. 

About the author Staff

Showcasing and curating a knowledge base of tech use cases from across the web.

TechForCXO Weekly Newsletter
TechForCXO Weekly Newsletter

TechForCXO - Our Newsletter Delivering Technology Use Case Insights Every Two Weeks