The PRESTO Innovation wins the SIMA gold award
This November at SIMA, ITK wins the first prize for innovation (SIMA gold award) within the “technology and services” category, with Presto, a tool to forecast tomato harvest and yield. It is a particularly significant milestone for ITK, given the importance of this exhibition for innovation in agriculture. Indeed, in 97 years, SIMA has rewarded more than 1,200 innovations including equipment, products, technologies and services presented by the tradeshow’s exhibitors.
Greenhouse tomato production : why forecast is important
Tomato is the most produced vegetable worldwide, with a volume of 187 MTons in 2020. It has many different varieties, that can be produced in different ways, in the field or in greenhouses.
Greenhouse production presents the advantage of extending the availability of fresh fruit in temperate countries like France, thanks to higher temperatures at the beginning and end of the growing season. It can also boost quality and yield, as plants in protected conditions are less exposed to insects and abiotic stresses.
For example, in France, 85% of tomato production occurs in greenhouses that are heated during the cold season. The increase in yield and availability that has resulted from greenhouse production has made tomato the most consumed fresh vegetable in France, a place that was previously held by potatoes.
This type of industrial production demands constant attention to logistics and market demands to maximise yield and profits. Greenhouse tomatoes have a continuous production over the growing season, with weekly harvests that are sold to supermarkets in pre-agreed quantities and prices. However, whereas greenhouse cultivation has generally increased yields, these can still fluctuate considerably from one week to the other due to a number of factors.
This can pose problems of both over-demand and over-production if the yield cannot be predicted accurately. Tomato price negotiation, which occurs in advance of harvest might not be favorable to growers, if yields are not correctly predicted. A wrong estimate of coming yields can also complicate logistics and agricultural operations. For example, if production is more abundant than expected, difficulties might be encountered at the moment of harvest if workforce or storage arrangements are lacking. According to the experts, in France errors cost up to 35% of total benefits, i.e. 20% of the sale price of sale of tomatoes.
For these reasons many tomato producers hire yield forecasters or use forecasting systems, that can through statistical and empirical methods, predict yields of the coming week. These predictions give producers a competitive advantage: if forecasted yields are lower than pre-agreed quantities, they can source additional produce from elsewhere, and if they are predicted to have excess fruits, they can look for alternative markets or arrange promotions. Ultimately a precise yield forecast smooths out all operation along the tomato supply chain and maximises benefits for growers.
Presto : the tool that allows you to organize your harvest and choose the best wholesale prices, two weeks in advance!
New technologies and statistical models are used more and more in modern smart agriculture to develop tools for crop management. To help tomato growers organize their operations and optimize their sales, ITK has developed a new, highly innovative forecasting tool, PRESTO.
PRESTO aims at perfectioning greenhouse tomato yield predictions using artificial intelligence algorithms that are trained on previous yield data. It contains a tomato harvest model that provides yield forecasts for the coming two weeks, only requiring weekly yield data as inputs.
PRESTO has been developed using historical yields and weather data from heirloom tomato varieties. These data have been used to train a random forest algorithm which has been showed to be as good or better than expert forecasts, with the additional advantage of forecasting up to two weeks in advance as opposed to only one week.
In addition, PRESTO is continuously perfectioning its predictions, thanks to a self-improving calibrating process, taking advantage of the yield data which are entered every week by users.
With PRESTO prediction errors are minimized and logistics and sales are made easier.
Agronomist and scientific editor