Automatic Face Aging in Videos via Deep Reinforcement Learning

Chi Nhan Duong1
Khoa Luu2
Kha Gia Quach1
Nghia Nguyen2
Eric Patterson3
Tien D. Bui1
Ngan Le4
1 Computer Science and Software Engineering, Concordia University, Canada
2 Computer Science and Computer Engineering, University of Arkansas, USA
3 School of Computing, Clemson University, USA
4 Electrical and Computer Engineering, Carnegie Mellon University, USA



Given a video, our approach can synthesize an aged-progressed video with consistent aging features across video frames.


This paper presents a novel approach for synthesizing automatically age-progressed facial images in video sequences using Deep Reinforcement Learning. The proposed method models facial structures and the longitudinal face-aging process of given subjects coherently across video frames. The approach is optimized using a long-term reward, Reinforcement Learning function with deep feature extraction from Deep Convolutional Neural Network. Unlike previous age-progression methods that are only able to synthesize an aged likeness of a face from a single input image, the proposed approach is capable of age-progressing facial likenesses in videos with consistently synthesized facial features across frames. In addition, the deep reinforcement learning method guarantees preservation of the visual identity of input faces after age-progression. Results on videos of our new collected aging face AGFW-v2 database demonstrate the advantages of the proposed solution in terms of both quality of age-progressed faces, temporal smoothness, and cross-age face verification.

Our proposed framework consists of three main processing steps: (1) Feature embedding; (2) Manifold traversal; and (3) Synthesizing final images from updated features. In the second step, a Deep Reinforcement Learning based framework is proposed to guarantee the consistency between video frames in terms of aging changes added during synthesis process.

Others Age-progressed Videos





Databases

The AginG Faces in the Wild (AGFW-v2) (Image Set) consists of 36299 images with the age ranging from 10 to 64 years old. This database is collected from three sources:

  1. The search engine using different keywords (i.e. "male 20 years old", etc.).
  2. The Productive Aging Laboratory (PAL) database [1].
  3. Mugshot images that are accessible from public domains.

Please notice that this dataset is made available for academic research purpose only. Images in the database are collected from the Internet, and the copyright belongs to the original owners. If any of the images belongs to you and you would like it removed, please kindly inform us, we will remove it from our dataset immediately.
The cropped faces of AGFW-v2 are available here

The AginG Faces in the Wild (AGFW-v2) (Video Set)

The video data and their metadata will be available soon ...

Reference
[1] M. Minear and D. C. Park. A life span database of adult facial stimuli. Behavior Research Methods, Instruments, & Computers, 36(4):630–633, 2004.

Paper and Citation

Automatic Face Aging in Videos via Deep Reinforcement Learning

Chi Nhan Duong, Khoa Luu, Kha Gia Quach, Nghia Nguyen, Eric Patterson, Tien D. Bui, Ngan Le

In CVPR, 2019  [arXiv]


Please add a reference if you are using the dataset. 

@INPROCEEDINGS{duong2019automatic,
author = {Automatic Face Aging in Videos via Deep Reinforcement Learning},
title = {Chi Nhan Duong and Khoa Luu and Kha Gia Quach and Nghia Nguyen and Eric Patterson and Tien D. Bui and Ngan Le},
booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, 
year={2019}
}

News and Tech Websites