Abstract
The present article analyses how data augmentation affects the performance of several deep learning models, including CNN, ResNet, and VGG19. By comparing performance indicators across different augmentation scenarios, this article assesses the effectiveness of augmentation strategies for each model. The results show that data augmentation greatly increased performance for the CNN model enhancing parameters like accuracy, precision etc. At the same time data augmentation had no influence on the ResNet model’s finel results although it showed a positive increasing in the process. Experiments also reveal very minor gains for the VGG19 model. These findings underscore the need of adapting augmentation strategies to each model's unique traits. Future research should concentrate on optimizing and investigating innovative data augmentation techniques to improve model performance and generalization.