On Recognizing Occluded Faces in the Wild

TL;DR

This paper introduces the ROF dataset with real-world occluded faces, highlighting the significant performance degradation of face recognition models on real occlusions compared to synthetic ones.

Contribution

The paper presents a new real-world occluded face dataset and evaluation protocols, revealing the limitations of current models on real occlusions.

Findings

Deep face models perform poorly on real occlusions.

Performance drops less on synthetically occluded faces.

The ROF dataset is publicly available for research.

Abstract

Facial appearance variations due to occlusion has been one of the main challenges for face recognition systems. To facilitate further research in this area, it is necessary and important to have occluded face datasets collected from real-world, as synthetically generated occluded faces cannot represent the nature of the problem. In this paper, we present the Real World Occluded Faces (ROF) dataset, that contains faces with both upper face occlusion, due to sunglasses, and lower face occlusion, due to masks. We propose two evaluation protocols for this dataset. Benchmark experiments on the dataset have shown that no matter how powerful the deep face representation models are, their performance degrades significantly when they are tested on real-world occluded faces. It is observed that the performance drop is far less when the models are tested on synthetically generated occluded faces. The ROF dataset and the associated evaluation protocols are publicly available at the following link https://github.com/ekremerakin/RealWorldOccludedFaces.