Today, different neural network models are used for different purposes. Some of these models have a high computational cost and it is not possible to train them with home or medium systems. Therefore, engineers and scientists use pre-trained models. In this research, I used CIFAR10 data. This dataset was published by the Canadian Institute for Advanced Research (CIFAR) and gained some of its popularity from the collaboration of Geoffrey Hinton and his collaborators. The CIFAR-10 dataset contains 60,000 32x32 color images in ten different classes and is used for training and testing various object recognition models.