Learning Disentangling and Fusing Networks for Face Completion Under Structured Occlusions

نویسندگان

  • Zhihang Li
  • Yibo Hu
  • Ran He
چکیده

Face completion aims to generate semantically new pixels for missing facial components. It is a challenging generative task due to large variations of face appearance. This paper studies generative face completion under structured occlusions. We treat the face completion and corruption as disentangling and fusing processes of clean faces and occlusions, and propose a jointly disentangling and fusing Generative Adversarial Network (DF-GAN). First, three domains are constructed, corresponding to the distributions of occluded faces, clean faces and structured occlusions. The disentangling and fusing processes are formulated as the transformations between the three domains. Then the disentangling and fusing networks are built to learn the transformations from unpaired data, where the encoderdecoder structure is adopted and allows DF-GAN to simulate structure occlusions by modifying the latent representations. Finally, the disentangling and fusing processes are unified into a dual learning framework along with an adversarial strategy. The proposed method is evaluated on Meshface verification problem. Experimental results on four Meshface databases demonstrate the effectiveness of our proposed method for the face completion under structured occlusions.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A 3D Dynamic Database for Unconstrained Face Recognition

In this paper, we present a new 3D dynamic face dataset dedicated to the development and test of algorithms which target face recognition under unconstrained conditions, from 3D videos. Several challenges which can occur in real world like scenarios are considered, such as continuous and freely-pose variation, expressive and talking faces, changes of the distance to the 3D camera, occlusions an...

متن کامل

Skin based Occlusion Detection and Face Recognition using Machine Learning Techniques

In this paper, a detailed experimental study of occlusion detection in the controlled environmentsbased on skin color is proposed.The image is given as an input to the face detection algorithm to detect the faces. Some faces are not detected dueto occlusion, so an occlusion detection technique is implemented to detect all the occluded faces. Those occlusions are detected using skin color of the...

متن کامل

Structured Tracking for Safety, Security, and Privacy: Algorithms for Fusing Noisy Estimates from Sensor, Robot, and Camera Networks

Structured Tracking for Safety, Security, and Privacy: Algorithms for Fusing Noisy Estimates from Sensor, Robot, and Camera Networks

متن کامل

Developing an Integrated Simulation Model of Bayesian-networks to Estimate the Completion Cost of a Project under Risk: Case Study on Phase 13 of South Pars Gas Field Development Projects

Objective: The aim of this paper is to propose a new approach to assess the aggregated impact of risks on the completion cost of a construction project. Such an aggregated impact includes the main impacts of risks as well as the impacts of interactions caused by dependencies among them. Methods: In this study, Monte Carlo simulation and Bayesian Networks methods are combined to present a frame...

متن کامل

Learning Pragmatics through Computer-Mediated Communication in Taiwan

This study investigated the effectiveness of explicit pragmatic instruction on the acquisition of requests by college-level English as Foreign Language (EFL) learners in Taiwan. The goal was to determine first whether the use of explicit pragmatic instruction had a positive effect on EFL learners’ pragmatic competence. Second, the relative effectiveness of presenting pragmatics through two deli...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/1712.04646  شماره 

صفحات  -

تاریخ انتشار 2017