Fecha de publicación: 07/04/2023 Fuente: Wipo "precision agriculture" This invention proposes a novel CNN-based neural network model named NPNet-19 that accurately classifies seven different kinds of maize crop diseases, including the healthy crop, with a higher level of classification accuracy when compared with other existing models. The proposed model was trained and tested using diverse datasets collected from public repositories and a real agriculture field, with various image augmentation techniques used to generate new images from the original ones, overcoming the challenge of limited image availability. A comparison of the proposed model with ten other existing models using several performance metrics, such as accuracy, loss, number of wrong predictions, precision, recall, and f1-score, demonstrated its superior classification performance. Accurately diagnosing crop diseases can lead to better yields and higher quality produce, ultimately improving the economic outcome of farming practices. The proposed model can reduce reliance on manual diagnosis, which can be time-consuming, expensive, and often subjective, by using a machine learning-based approach. It can also increase access to diagnosis, particularly in regions where expert diagnosis is not readily available or affordable, ultimately improving the management of crop diseases and reducing the impact of diseases on crop yields and food security. The invention has the potential to foster innovation in precision agriculture by expanding the use of machine learning techniques in the field of crop disease diagnosis, leading to new advancements in precision agriculture and improving the sustainability and efficiency of food production.