Modelo de machine learning aplicado à identificação do mosquito Aedes aegypti

Abstract

The present project aimed to develop and train a machine learning model for the classification of Aedes aegypti mosquitoes and other mosquito species, with the purpose of contributing to the control of diseases transmitted by these vectors. The development process of the model included the collection of mosquito image data, the implementation of preprocessing techniques for effective model training, and the implementation of a convolutional neural network for sample classification. The dataset consisted of 14,716 images of mosquitoes from various species, with an imbalance between the classes. Data Augmentation technique was applied to increase the number of samples and balance the training set. The model was trained for 60 epochs, with the use of callbacks to optimize the training process. The results showed an accuracy of 75.4% in classifying the samples, indicating a promising performance of the model. However, signs of possible overfitting were observed due to the lack of representative samples and limitations in computational resources. Despite the challenges faced, the model demonstrated the ability to distinguish between mosquito species, contributing to advancements in the control of Aedes aegypti and other mosquito species. Future work may explore expanding the dataset and experimenting with different neural network architectures to improve the model's performance. In summary, the project represents an important step in the development of machine learningbased solutions for the prevention of mosquito-borne diseases.