Proven 90% accuracy rose yield prediction by drones
Added on 28 October 2024
The rose production and retail demand misalignment costs are conservatively estimated at 50 million euro worldwide (5-10% of the total rose trade). Aligning rose production and demand is a daily challenge for rose growers delivering directly to retailers. Manually counting bud- and flower stages costs time and manual counts are not objective. A 20% average (up to 50% for certain varieties) daily mismatch in prediction and realized production is common for many rose growers.
Corvus Drones and GrowerAdviser are launching the rose yield prediction application. The drone flies fully automatic in the greenhouse, takes pictures of the crop which are analyzed by AI powered computer vision software. Reporting is via an interactive web application or email alert. The yield prediction is offered as a service and is easy to operate. The Drone as a Service is used by growers in more than 15 countries like The Netherlands, Germany, Ecuador, USA or Kenya.
At several Dutch and Ecuadorian rose growers the yield prediction application has been tested and validated. After one year an amazing accuracy level of 90% has been proven for a multi-day prediction and 85% accuracy for a multi-week prediction. The following varieties are validated: Red Naomi, Avalanche, Cotopaxi, Sweetness, Tycoon and Deep Purple. Human count and prediction has been beaten by the drone and software.
This 90% high accuracy level is achieved by combining bud and flower stage counts with greenhouse climate data, historical data and weather forecast. GrowerAdviser, a subsidiary of Quantum_Inc., is taking care of the data science part. These analytics are fully incorporated in the Corvus Drones reporting. Growers by the Drone as a Service and do not need to invest. The drone is capable of flying in any greenhouse at any place in the world if there is 250cm of flight space.
The 3 companies are also testing the powdery mildew detection. Powdery mildew is one of the most harmful diseases in roses worldwide and costs growers lots of money. The results are above expectations, because small spots of 1-2 mm are automatically detected and reported. At an interactive greenhouse map, the infestations are clearly indicated for swift actions by the growers.
The rose yield prediction application will be launched at the IFTF exhibition (booth A1.27), November 5-7 in The Netherlands.
More information:
Corvus Drones
Frans-Peter Dechering
GrowerAdviser
Tatiana Nesterenko
More news
