CHALLENGE

Weed growth in grain crops – especially soybeans – has a considerable negative impact on productivity. Unfortunately, when weeds become easily visible, it is an indication that they have already spread. Thus, a model of artificial intelligence of computer vision for early identification of the problem could provide an important competitive advantage.

DEVELOPMENT

CERTI trained artificial intelligence models using images of soybean fields taken by the client company’s drones. The data was labeled by specialized agronomists and served as input for the training process. The process involved researching and employing advanced techniques of deep learning with convolutional neural networks over nine months.

RESULT

CERTI delivered not only a trained model capable of identifying weeds in soybeans at different crop stages, but also an AI framework that enables the company to train its own models, including for other crops and other elements found in the crops.