RMSE Cooperative: 1.43RMSE Indoor: 1.42RMSE Outdoor: 1.42
Papers with Code · CC BY-SA 4.0
3D Face Reconstruction
3D Face Reconstruction is a computer vision task that involves creating a 3D model of a human face from a 2D image or a set of images. The goal of 3D face reconstruction is to reconstruct a digital 3D representation of a person's face, which can be used for various applications such as animation, virtual reality, and biometric identification.
<span style="color:grey; opacity: 0.6">( Image credit: 3DDFA_V2 )</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
HiFace-f
@nose: 1.036 (±0.280)@mouth: 1.450 (±0.413)@forehead: 1.324 (±0.334)@cheek: 1.291 (±0.362)
논문
-
DenseLandmarks (Multi-view)
Mean Reconstruction Error (mm): 1.01Stdev Reconstruction Error (mm): 0.84Median Reconstruction Error: 0.81
논문
-
데이터셋
9개논문
RMSE Cooperative: 1.60RMSE Indoor: 1.61
Average 3D Error: 1.82
RMSE Cooperative: 1.34RMSE Indoor: 1.53
Mean NME : 6.3833%
Average 3D Error: 0.95
Mean NME : 3.7551%
Mean NME : 5.2667%
Average 3D Error: 1.93
Average 3D Error: 1.50
라이브러리
Papers with Code 리더보드 row에 연결된 구현 저장소입니다.