The Devil of Face Recognition is in the Noise

Fei Wang, Liren Chen, Cheng Li, Shiyao Huang, Yanjie Chen, Chen Qian, Chen Change Loy ; The European Conference on Computer Vision (ECCV), 2018, pp. 765-780

Abstract


The growing scale of face recognition datasets empowers us to train strong convolutional networks for face recognition. While a variety of architectures and loss functions have been devised, we still have a limited understanding of the source and consequence of label noise inherent in existing datasets. We make the following contributions: 1) We contribute cleaned subsets of popular face databases, i.e., MegaFace and MS-Celeb-1M datasets, and build a new large-scale noise-controlled IMDb-Face dataset. 2) With the original datasets and cleaned subsets, we profile and analyze label noise properties of MegaFace and MS-Celeb-1M. We show that a few orders more samples are needed to achieve the same accuracy yielded by a clean subset. 3) We study the association between different types of noise, i.e., label flips and outliers, with the accuracy of face recognition models. 4) We investigate ways to improve data cleanliness, including a comprehensive user study on the influence of data labeling strategies to annotation accuracy. The IMDb-Face dataset has been released on https://github.com/fwang91/IMDb-Face.

Related Material


[pdf]
[bibtex]
@InProceedings{Wang_2018_ECCV,
author = {Wang, Fei and Chen, Liren and Li, Cheng and Huang, Shiyao and Chen, Yanjie and Qian, Chen and Change Loy, Chen},
title = {The Devil of Face Recognition is in the Noise},
booktitle = {The European Conference on Computer Vision (ECCV)},
month = {September},
year = {2018}
}