CROP CLASSIFICATION AND UNWANTED PLANT DETECTION USING ARTIFICIAL INTELLIGENCE IN AERIAL IMAGERY FOR PRECISION AGRICULTURE

In light of the paramount importance of precision agriculture, this research proposes an algorithm leveraging artificial intelligence that can classify various crop types and identifying the presence of weeds within the designated study areas. The present algorithm was developed in Python and incorporates convolutional neural network (CNN) models that were specifically designed to segment regions with carrot and potato crops. Segmentation is achieved through the implementation of a U-shaped neural model, which utilizes data acquired from aerial imagery. This information undergoes a process of analysis and selection, with the objective of subsequent integration into the network. Furthermore, a classification model is employed that emphasizes the differentiation between the two crop types. In conclusion, the CNN response is regarded as a novel estimate, which, in this instance, is directed towards the identification of areas that are afflicted with weeds. The final process is associated with a recently developed semantic segmentation model that is also U-shaped. This model has been adapted to identify weeds in the designated fields.