A Light CNN for Deep Face Representation with Noisy Labels

9 Nov 2015Xiang WuRan HeZhenan SunTieniu Tan

The volume of convolutional neural network (CNN) models proposed for face recognition has been continuously growing larger to better fit large amount of training data. When training data are obtained from internet, the labels are likely to be ambiguous and inaccurate... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Age-Invariant Face Recognition CACDVS MFM-CNN Accuracy 97.95% # 5
Age-Invariant Face Recognition CAFR Light CNN Accuracy 73.56% # 2
Face Verification Labeled Faces in the Wild Light CNN-29 Accuracy 99.33% # 7
Face Verification MegaFace Light CNN-29 Accuracy 85.133% # 6
Face Identification MegaFace Light CNN-29 Accuracy 73.749% # 4
Face Verification YouTube Faces DB Light CNN-29 Accuracy 95.54% # 5