AgroPages-Bayer launches WeedScout in Brazil- first application to identify weeds through cell images-Agricultural news

Bayer announced this month in Brazil the launch of the WeedScout application, which recognizes weeds through images taken by cell phones. The goal, according to the company, is to allow WeedScout to identify weeds during their initial stages, when controlling the plant is easier.

“Managing the properties, optimizing the monitoring, and allowing a more precise and fast decision-making by the producer, with a reduction of costs, increasing yields and greater assertiveness, is the main goal of the app launched by Bayer,” according to the Digital Farming Director in Brazil, André Salvador.

Also, the multinational repeated that it intends to put in place the concept of “Digital Farming” to help to combat resistant weeds that are difficult to control. With WeedScout, farmers can monitor the competition between plants with necessary information to use the most appropriate pesticides. With over 30,000 images registered, the application relies on a collaborative update system, which is compiled through the use of photos taken by users to enhance algorithms.

“The weed is a threat that can hurt any crop. So it is indispensable that the application database be large. As more images are included, more plants will be identified and with greater precision. This is what will increase the accuracy in the identification of species that are in WeedScout, along with the capacity to recognize new species. And we want it to be done in a collaborative way,” noted André Felli, Digital Farming Business Development Manager at Bayer.

A result of a global project, this application – available for Android and iOS - has algorithms to allow for the recognition of the major species of weeds. “In order to form this database, over 30,000 photos were taken around the world,” said Felli. WeedScout works based upon the concept of MVP, which stands for MinimumViableProduct, meaning that even after the app is offered, it continues to be improved as it is put into use.

“We have a team of researchers that specialized in the issue that make the validation and enhancement of algorithms, based in the database, conform to new images that are generated,” he added.