DESENVOLVIMENTO DE ALGORITMO DE INTELIGÊNCIA ARTIFICIAL PARA DETECÇÃO DE FALHAS DE PLANTIO : UMA APLICAÇÃO AO CULTIVO DE MILHO

Abstract

This ѕtᴜdу рrороѕеѕ the dеѕіgn оf an algorithm anchored in Artificial Intelligence, aiming to identify gaps in the planting process of energy crops, especially focusing on corn production.To achieve this, the Vision Transformer (ViT) architecture was employed, accompanied by transfer learning techniques, to train a model capable of classifying visual records captured by drones into three classificatory groups: plant, soil, and straw.The dataset was structured from authentic images of fields, provided by partners in the agricultural sector, and the training was carried out on the Google Colab platform, utilizing specialized libraries such as PyTorch and Hugging Face.The performances obtained evidenced by metrics such as accuracy and the F1-score indicate that the model proved effective in identifying irregularities in the planted spacing.Analyzing the results, a promising potential of AI is perceived as a strategic tool for intelligent management in precision agriculture, boosting operational efficiency and well-founded decision-making in the field.