Pest and disease are two main enemies for the agriculture growers and agrifood industry. A potential solution to tackle these issues is to introduce precision farming technologies through a pest and disease detection model that will allow the grower/industrial user to detect such crop enemies early on, thus reducing pesticide applications and optimizing the productivity. This proposed solution will be implemented in this use case by addressing the tomato crop to optimize its productivity. Previous studies have already produced models, applications or software that warn farmers against the occurrence of an attack, but there are excessive false positive events, due to micro-weather circumstances or other reasons. By adding image analysis resorting to ANNs (something not present in previous tools) we expect to reduce the number of false positives. These false positives lead to unnecessary treatments, impacting both the environment and the crop expenditure. In addition, there is a need for quick action. The system shall be able to raise warnings quickly to allow a prompt action from the farmer to deal with the crop enemy.
Inov, CCTI, Italagro, TPRO-Technologies, CEA and STGNB team together to implement AI techniques on the data coming from the environmental sensors to perform data analysis at the edge to achieve the following outcomes: a binary classification model that detects if there are any pests or diseases attacking the crop, a multiclass classification model that distinguishes between 5 different pests and diseases, identify correlations and patterns between meteorological data and the presence of pests and diseases, and finally decision-making actions in near real-time at system level.