mean_variance_loss_for_deep_age_estimation_from_a_face.org

Paper Information

Title

Mean-Variance Loss for Deep Age Estimation from a Face

Author

Hongyu Pan, Hu Han, Shiguang Shan and Xilin Chen

Published

CVPR 2018

Code

None

Paper Content

Why

age regression or classification do not utilize the distrubutions robustness or ordinal information in age

How

Use 3 loss

1. original cross_entropy loss
2. mean loss, softmax output * range_j
3. variance loss, shrinkt probabilities

Results

Test in Morph II and FGNet

Thoughts

1. softmax expectation from deep expection penalize ages differently
2. why use alexnet and vgg in 2018?