The construction of an Artificial Neural Network model for image recognition requires a high degree of specialization and simulations to be able to configure all the parameters necessary for a good model performance. This paper presents a method for analyzing the main configuration hyperparameters of a Convolutional Image Recognition Neural Network to make the task of choosing configuration parameters more accessible and can be extended to other neural network models that use them. parameters, thus shortening the model construction work. Going through the main configuration parameters, as the ideal ranges of values are defined, the method leads to the following parameters, allowing the improvement of the neural network result and experimentation with the parameters of the previous steps until the ideal result is reached. The process proved to be effective, since, starting from a previously tested model, the optimizations were removed and the method tests were applied, reaching a more optimized and better performing final configuration than the original model.
