.png){kind=icon}
Return
CED-Net: contextual encoder-decoder network for 3D face reconstruction
.png){kind=icon}
PRE
AI
DOI：10.1007/s00530-022-00938-2.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a contextual encoder-decoder network to regress UV position map for 3D face reconstruction, named CED-Net. Specifically, the CED-Net introduces contextual information both in the shape and feature level, which is served as the crucial information. Firstly, shape context relationship is considered as the loss function constraint, in which the intrinsic Euclidean norm and vector angle similarity are computed for each contextual vector. Furthermore, to add the contextual information at the feature level, the local feature correlation modulator is incorporated into the network's center section, which directs the network to capture the relationship between facial features from a spatial perspective. The heatmaps of the learned features demonstrate that the proposed CED-Net can pay attention to the comprehensive facial regions for reconstruction. The quantitative and qualitative experiments conducted on AFLW2000-3D and AFLW-LPFA show that the proposed method achieves superior results in both benchmarks.
Keywords:
3D face reconstruction
Encoder-decoder network
Shape context
Local correlation
Journal
IF:
3.1
Papers:
2.8K
Citations:
2.7K
Organization
No organization information available
Cited Papers
ShareSaveShareSaveShareSaveShareSaveShareSaveUse of NMR as an analytical tool in the process development of conjugate vaccines against Haemophilus influenzae type b (Hib) and meningococcal serogroup A (MenA).png)
Biologicals
BiologicalsIF0
PREAI
ShareSaveVideo compression using hybrid hexagon search and teaching-learning-based optimization technique for 3D reconstructionMULTIMEDIA SYSTEMS
MULTIMEDIA SYSTEMSIF3.1
PREAI
ShareSave
View more