An image analysis service for agriculture

Publish on 29/01/2020
Are the apricots ripe? Are insects threatening the squash? Have all the cows had their share of fodder? All questions that require the farmer or rancher to make rounds. The fruit of research into robotic vision at the Inria Rennes – Bretagne Atlantique centre, one startup, Dilepix, is getting ready to offer an online image analysis service to automate monitoring and help save valuable time.
Illustration agriculture
Photo Mickael Tournier via Unsplash, CC0

In breton, Lagadic means “little eyes ”. It is also the name of a scientific team whose algorithms give vision to medical instruments, drones, wheelchairs and spacecraft. The results of its research will soon be applied to agriculture.

Engineer on the team for five years, Aurélien Yol  is putting the finishing touches to a business creation project supported by InriaHub , a programme for developing startups spun off from research carried out at Inria.
“Why am I interested in agriculture? It’s a sector that is rapidly making the digital transition for which our vision technologies have a great deal to offer. Video cameras are already being used to monitor livestock. It ensures that each animal can properly access its share of feed in the stall area. Monitoring systems are also starting to be used in the fields to observe the condition of crops and detect pests. For example, we have a partner who sells video cameras located in insect traps. These devices are equipped with a network card. The images are transmitted to the farmer’s computer so that he no longer needs to make rounds. It’s a real time saver. ” It’s no longer necessary to physically check the fields every day.
“Farmers would also like to save time spent watching the screens to know if there are problems to resolve. That's how our technologies can help them. We are going to provide an automatic image analysis system that will keep the user informed.  It may involve sending a message and quantifying the number of insects in the traps, as well as ensuring that the livestock is feeding correctly or that fruit is ripe for harvesting. We use algorithms that learn from the data. We can thus adapt the tool to the specific nature of the various crops and their special needs. ”

For farmers and manufacturers of farm equipment

This image analysis module is intended for farmers, agricultural cooperatives and manufacturers of farm equipment.  “It can supplement the service already offered by the designers of hardware solutions. One possibility we’re considering is embedded analysis. In insect traps, for example, we can include an additional card or use the card that operates the video camera. ” Aurélien Yol is currently working on preparing a proof of concept with an equipment manufacturer. “It’s about setting up an analysis service for the first real field trial in order to be able to test this technology in real conditions. ” 
Also under study: defining the business model.  “Drone services are paid based on the surface area that is monitored. The designers of connected video cameras bill instead by the number of machines installed. Another way of operating consists in taking into account the volume of images analysed, knowing that monitoring livestock is an ongoing effort while crops only require images at certain times. ” One also has to take into account that “most of these activities are highly seasonal ”. Speaking of calendars, what’s coming up for the company? “It should be formed by the end of the year. I now have an associate, Alban Pobla , who will focus more on sales. We’ll probably start raising capital next spring. ”