Analyzing the effectiveness of image augmentations for face recognition from limited data

TL;DR

This paper evaluates various image augmentation techniques, including basic manipulations and generative methods, demonstrating that their combination significantly enhances face recognition accuracy from limited data.

Contribution

It provides a comprehensive analysis of augmentation strategies and shows that combining generative and basic methods yields superior face recognition performance.

Findings

Augmentations improve face recognition accuracy.

Combination of generative and basic augmentations outperforms individual methods.

Augmentation techniques are effective even with limited data.

Abstract

This work presents an analysis of the efficiency of image augmentations for the face recognition problem from limited data. We considered basic manipulations, generative methods, and their combinations for augmentations. Our results show that augmentations, in general, can considerably improve the quality of face recognition systems and the combination of generative and basic approaches performs better than the other tested techniques.