RECORD
LI-Net: Large-Pose Identity-Preserving Face Reenactment Network
Computer Vision and Pattern Recognition · Submitted 7 Apr 2021
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim Both qualitative and quantitative experimental results demonstrate the superiority of our method.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Face reenactment is a challenging task, as it is difficult to maintain accurate expression, pose and identity simultaneously. Most existing methods directly apply driving facial landmarks to reenact source faces and ignore the intrinsic gap between two identities, resulting in the identity mismatch issue. Besides, they neglect the entanglement of expression and pose features when encoding driving faces, leading to inaccurate expressions and visual artifacts on large-pose reenacted faces. To address these problems, we propose a Large-pose Identity-preserving face reenactment network, LI-Net. Specifically, the Landmark Transformer is adopted to adjust driving landmark images, which aims to narrow the identity gap between driving and source landmark images. Then the Face Rotation Module and the Expression Enhancing Generator decouple the transformed landmark image into pose and expression features, and reenact those attributes separately to generate identity-preserving faces with accurate expressions and poses. Both qualitative and quantitative experimental results demonstrate the superiority of our method.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Face reenactment is a challenging task, as it is difficult to maintain accurate expression, pose and identity simultaneously. | background 1.00 | 0.07 | 0.03 | asserted 1.00 |
| s1 | Most existing methods directly apply driving facial landmarks to reenact source faces and ignore the intrinsic gap between two identities, resulting in the identity mismatch issue. | background 0.73 | 0.04 | 0.03 | asserted 1.00 |
| s2 | Besides, they neglect the entanglement of expression and pose features when encoding driving faces, leading to inaccurate expressions and visual artifacts on large-pose reenacted faces. | background 0.82 | 0.05 | 0.02 | asserted 1.00 |
| s3 | To address these problems, we propose a Large-pose Identity-preserving face reenactment network, LI-Net. | contribution 1.00 | 0.82 | 0.09 | asserted 0.95 |
| s4 | Specifically, the Landmark Transformer is adopted to adjust driving landmark images, which aims to narrow the identity gap between driving and source landmark images. | method 0.98 | 0.83 | 0.04 | asserted 0.97 |
| s5 | Then the Face Rotation Module and the Expression Enhancing Generator decouple the transformed landmark image into pose and expression features, and reenact those attributes separately to generate identity-preserving faces with accurate expressions and poses. | method 0.99 | 0.87 | 0.09 | asserted 0.99 |
| s6 | Both qualitative and quantitative experimental results demonstrate the superiority of our method. | result 0.96 | 0.93 | 0.81 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Face reenactment is a challenging task, as it is difficult to maintain accurate expression, pose and identity simultaneously. Most existing methods directly apply driving facial landmarks to reenact source faces and ignore the intrinsic gap between two identities, resulting in the identity mismatch issue. Besides, they neglect the entanglement of expression and pose features when encoding driving faces, leading to inaccurate expressions and visual artifacts on large-pose reenacted faces. To address these problems, we propose a Large-pose Identity-preserving face reenactment network, LI-Net. Specifically, the Landmark Transformer is adopted to adjust driving landmark images, which aims to narrow the identity gap between driving and source landmark images. Then the Face Rotation Module and the Expression Enhancing Generator decouple the transformed landmark image into pose and expression features, and reenact those attributes separately to generate identity-preserving faces with accurate expressions and poses. Both qualitative and quantitative experimental results demonstrate the superiority of our method.
Human review
How the literature uses it
What this paper does with the arXiv papers it cites
What it cites
Uses the method or tool of 2 papers
-
In order to constrain that reenacted images $Î_s,q$ share same semantic and identity information as $I_s,q$, two perceptual networks are utilized, which are VGG-19 [ref] for image classification and VGGFace [ref] for face verification.
Very Deep Convolutional Networks for Large-Scale Image Recognition · Expression Enhancing Generator · 1.00
-
Three modules in LI-Net are trained separately using the Adam optimizer [ref] .
Adam: A Method for Stochastic Optimization · Experimental Settings · 1.00
Cites as background 1 paper
-
Diverse methods related to GAN [ref] are proposed to achieve facial attribute editing or whole face image generation. cGAN [ref] controls the attribute of faces using condition information vectors.
Conditional Generative Adversarial Nets · Related work · 1.00
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.