Skip to Main content Skip to Navigation
Conference papers

Monocular 3D Head Reconstruction via Prediction and Integration of Normal Vector Field

Abstract : Reconstructing the geometric structure of a face from a single input image is a challenging active research area in computer vision. In this paper, we present a novel method for reconstructing 3D heads from an input image using a hybrid approach based on learning and geometric techniques. We introduce a deep neural network trained on synthetic data only, which predicts the map of normal vectors of the face surface from a single photo. Afterward, using the network output we recover the 3D facial geometry by means of weighted least squares. Through qualitative and quantitative evaluation tests, we show the accuracy and robustness of our proposed method. Our method does not require accurate alignment due to the image-to-image translation network and also successfully recovers 3D geometry for real images, despite the fact that the model was trained only on synthetic data.
Document type :
Conference papers
Complete list of metadata

Cited literature [50 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-02928494
Contributor : Mohamed Daoudi Connect in order to contact the contributor
Submitted on : Wednesday, September 2, 2020 - 3:40:25 PM
Last modification on : Tuesday, January 4, 2022 - 6:12:58 AM
Long-term archiving on: : Wednesday, December 2, 2020 - 4:53:23 PM

File

VISAPP2020-2.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02928494, version 1

Citation

Oussema Bouafif, Bogdan Khomutenko, Mohamed Daoudi. Monocular 3D Head Reconstruction via Prediction and Integration of Normal Vector Field. 15th International Conference on Computer Vision, Theory and Applications., Feb 2020, Valletta, Malta. ⟨hal-02928494⟩

Share

Metrics

Les métriques sont temporairement indisponibles