This article examines the influence of normalization and augmentation methods on the accuracy of face recognition and facial expression recognition. It is shown that the performance of computer vision algorithms depends not only on the model architecture, but also on the quality of image preprocessing. The main factors affecting classification accuracy include variations in illumination, scale, pose, head orientation, image quality and training data structure. The paper analyzes the main normalization methods, including face alignment, image resizing, brightness and contrast normalization, as well as augmentation techniques such as rotations, horizontal flipping, brightness variation, noise injection and random geometric transformations. Their role in improving model robustness to input variability and reducing overfitting is discussed. The analysis shows that the combined use of normalization and augmentation improves the quality of both identity recognition and facial expression classification. The obtained conclusions confirm the effectiveness of these methods as an essential component of modern face-based classification systems.
face recognition
facial expression recognition
normalization
data augmentation
computer vision
deep learning
classification