H2020 AGERPIX Project: Artificial intelligence for yield estimations in fruit orchards
- Type Project
- Status Filled
- Execution 2019 -2020
- Assigned Budget 50.000,00 €
- Scope Europeo
- Main source of financing H2020
- Project website AGERPIX
Crop yield estimation is a central function in fruit production, as it determines which portion of the crop is suitable for consumption and marketing. Yield estimation, which is crucial for apple production, must be reliable, rapid, and cost-effective. Today, fruit is counted manually by farm workers. This is not only time-consuming, but accuracy cannot be guaranteed, especially for large yields.
The EU-funded AGERPIX project is developing a solution based on artificial intelligence algorithms. Its goal is to offer reliable and rapid predictions, significantly reducing operating and management costs. It provides producers with information on fruit size and quality, as well as production yields. It calculates the required labor and offers logistics planning.
Crop yield estimation is an important task in apple orchard management. Current yield estimation practice relies on manual fruit counting by workers. It is time-consuming and labor-intensive, highly inaccurate, and impractical for large fields.
Agerpix provides accurate predictions to help growers improve fruit quality and reduce operating costs by making better decisions about fruit thinning intensity and plant nutrients and treatments (mid-season), harvesting workforce size, machinery and materials, and logistical planning for storage, packing, and cold storage. This includes developing a marketing strategy tailored to expected production, achieving a 50% cost reduction in orchard management operations. Artificial Intelligence algorithms are used to identify and measure diameter ranges and visualize fruit bushiness and vigor, providing yield estimates based on plant height and plant health variables.
Several pilot projects for the yield estimation system have been implemented at major apple producers (Nurfri, Spain's No. 1, and Blue Whale, France's No. 1, among others) with over 140 hectares analyzed with an accuracy of 90–95%. The AGERPIX system has been adapted to four different apple varieties. Following successful validation activities, CODESIAN is developing a client portfolio worth €38 million over five years.
However, because AGERPIX is offered as a B2B service and because the technology can be easily replicated on other fruits (validations underway with table grapes and peaches with minor AI/sensor adaptations), a careful scale-up design is needed to strengthen the business plan to cover the global needs of fruit markets (apple: €517 million; table grape: €124 million; peach: €160 million; mandarin: €299 million; avocado: €54 million). After gathering more data through extensive validations of new fruits, CODESIAN projects +€7.9 million in revenue with +€4.5 million in EBIT and +60 new jobs created by 2024.
Tree Fruit Detection System Aids Precision Farming The high-precision fruit counting system uses artificial intelligence to help farmers optimize harvests and plan market sales. Farmers need to know their expected yields so they can plan accordingly. But counting fruit is a time-consuming task. The EU-funded AGERPIX project has devised a more efficient way for agricultural suppliers: an automated tree fruit counting system. It allows growers to assess the quantity of fruit they will grow with accuracy levels of up to 95%.
The system developed by AGERPIX allows farmers to assess the orchard tree by tree, determines plant vigor, provides a leafiness index, and lets them know exactly which areas are most productive. Through the use of cameras, precision positioning, and artificial intelligence (AI), it enables farmers to gather the true production values of their farms. The system offers multiple benefits, including orchard management, logistics fleet organization, optimization of cold storage equipment, shipping, and human resources, improved data processing and costs, and perhaps most importantly, improved fruit quality.
"In short, the goal is to make plantations produce more, better, and with fewer resources," says David Frances, CEO of Agerpix Technologies, the project host. The roving detector system travels through the plantations, row by row, with a series of sensors carried on a quad bike. The data received through the system is analyzed and processed with AI to identify the separated fruit on each tree. The results are then delivered to the producer and can be accessed from any device. "AI is a very applicable technology to crops; it allows us to generate harvest predictions, fruit phenological status, plant vigor, and potential pests," notes Frances, AGERPIX project coordinator. The system is able to identify the plant's physiological state, allowing it to detect nutrient deficiencies, water stress, and diseases. All production data, records, and analysis are stored securely and privately in the cloud. A "night operations" mode allows the system to operate in all conditions.
The high-performance, energy-efficient LED lighting means it can operate at night without interfering with other farming tasks. The integrated GPS component provides highly accurate georeferenced data, separating plots of land according to their corresponding production values.
This is an incredibly useful tool for farmers to see where their land's strengths lie. The product is currently offered as a service, but the company plans to develop fully autonomous data collection systems in the future. Fueling the Future "With society's growing demand for food, population growth, the problem of water insufficiency, and the impossibility of new crop fields, increasing plantation productivity is critical, and that's our fundamental goal," Frances explains. The EU grant helped the team generate and focus their business plan, which they are now moving forward to implement.
Agerpix will continue to generate value through its information, Frances adds: "To do this, we are developing new ways to help producers make better decisions." The company is further developing the technology to increase the number of species and varieties the system can detect. "We continue to develop other types of products related to precision agriculture, solving problems raised by producers themselves," says Frances.
- CODESIAN SOFTWARE TECH SL (CODESIAN SOFTWARE TECH)