Agriculture is vital to global development and economic prosperity. In recent years, the use of Machine learning has shown great potential for the classification of pest images. The objective is to compare the performance of various configurations of a Machine Learning model for the classification of pest images in olive oil plantations in Tacna, Peru. The generation of dataset for the three types of pests was obtained using a digital camera. The VGG16 architecture was used to compare models by altering the input size and a comparison was made with and without the application of data augmentation. The results showed significant differences in model performance, evidencing that settings such as the size of the input images and the application of data augmentation techniques have a considerable impact on the classification accuracy. The input dimensions 299x299 and 452x452 achieved the highest accuracy (100% maximum, 99.27% average) in the classification of olive pests without data augmentation. Furthermore, performance was significantly affected by the input dimensions and the use of this technique, as confirmed by the statistical analysis with p values (0.001) <0.05.