Papers with Code · CC BY-SA 4.0

Face Generation

3D Face Reconstruction › Facial Recognition and Modelling › Face Generation 24 평가 기록 12 datasets

Face generation is the task of generating (or interpolating) new faces from an existing dataset.

The state-of-the-art results for this task are located in the Image Generation parent.

<span style="color:grey; opacity: 0.6">( Image credit: Progressive Growing of GANs for Improved Quality, Stability, and Variation )</span>

개요는 각 항목의 일부입니다. 위 탭에서 전체 기록을 페이지별로 탐색할 수 있습니다.

벤치마크

Papers with Code 아카이브의 데이터셋별 기록 예시입니다. 평가 조건과 지표 방향이 검증되지 않아 순위를 의미하지 않습니다.

Dataset기록된 모델 예시MetricPaperCode

VoxCeleb2 - 1-shot learning

Fast Bi-layer Avatars (medium size)

LPIPS: 0.358SSIM: 0.508CSIM: 0.653Normalized Pose Error: 43.3

논문 코드

LRS2

Wav2Lip + ViT + MARLIN

LSE-D: 7.127LSE-C: 5.528FID: 3.452

논문 코드

LRS3

Wav2Lip + GAN

LSE-D: 6.986LSE-C: 7.574FID: 4.35

논문 코드

LRW

Wav2Lip + GAN

LSE-D: 6.774LSE-C: 7.263FID: 2.475

논문 코드

VoxCeleb1 - 1-shot learning

Few-shot Adversarial Model

FID: 43.0

논문 코드

VoxCeleb1 - 32-shot learning

Few-shot Adversarial Model

FID: 29.5

논문 코드

VoxCeleb1 - 8-shot learning

Few-shot Adversarial Model

FID: 38.0

논문 코드

VoxCeleb2 - 8-shot learning

CainGAN

FID: 24.9

논문 -

100 sleep nights of 8 caregivers

Ashok

10%: 12

논문 코드

CREMA-D

EmoGen

EmoAcc: 83.2FID: 5.29LSE-C: 6.663

논문 코드

LRW

LipGAN

LMD: 0.60SSIM: 0.96

논문 코드

VoxCeleb2 - 32-shot learning

Few-shot Adversarial Model

FID: 30.6

논문 코드

데이터셋

12개

VoxCeleb2 - 1-shot learning LRS2 LRS3 LRW VoxCeleb1 - 1-shot learning VoxCeleb1 - 32-shot learning VoxCeleb1 - 8-shot learning VoxCeleb2 - 8-shot learning 100 sleep nights of 8 caregivers CREMA-D LRW VoxCeleb2 - 32-shot learning

논문

Few-Shot Adversarial Learning of Realistic Neural Talking Head Models

1905.08233 2019-05-20 이 페이지의 평가 4건

FID: 42.2

Fast Bi-layer Neural Synthesis of One-Shot Realistic Head Avatars

2008.10174 2020-08-24 이 페이지의 평가 3건

LPIPS: 0.358SSIM: 0.508CSIM: 0.653Normalized Pose Error: 43.3

Pose Manipulation with Identity Preservation

2004.09169 2020-04-20 이 페이지의 평가 2건

FID: 24.9

X2Face: A network for controlling face generation using images, audio, and pose codes

2018-09-01 이 페이지의 평가 2건

FID: 51.5

0-Step Capturability, Motion Decomposition and Global Feedback Control of the 3D Variable Height-Inverted Pendulum

1912.06078 2019-12-12 이 페이지의 평가 1건

10%: 12

라이브러리

Papers with Code 리더보드 row에 연결된 구현 저장소입니다.

github.com/grey-eye/talking-heads 1 논문

github.com/Ierezell/PapierFewShot 1 논문

github.com/shoutOutYangJie/Few-Shot-Adversarial-Learning-for-face-swap 1 논문

github.com/vincent-thevenin/Realistic-Neural-Talking-Head-Models 1 논문

github.com/ZVK/talking_heads 1 논문

github.com/ZVK/Talking-Heads 1 논문

github.com/saic-violet/bilayer-model 1 논문

github.com/GabrielEGC/IHMC-Robotics 1 논문

하위 태스크

TH Talking Head Generation 10 데이터셋 · 22 rows TF Talking Face Generation 2 데이터셋 · 2 rows FA Face Age Editing 0 데이터셋 · 0 rows FE Facial expression generation 0 데이터셋 · 0 rows KF Kinship face generation 0 데이터셋 · 0 rows