VN3DFace: 3D facial expression database of Vietnamese

The availability of 3D face datasets has surged, enhancing computer vision technologies and enabling better facial recognition systems. However, unique facial features across different ethnic groups necessitate tailored datasets. Despite their importance, creating 3D facial data is resource-intensive, requiring advanced hardware and expertise. This paper presents a solution using 3D face reconstruction techniques to develop a 3D emotional face dataset from images of Vietnamese individuals, named VN3DFace. This approach can be applied to any 2D facial image dataset, facilitating the generation of larger, diverse datasets. VN3DFace demonstrates promising results, achieving an LPIPS score of 0.09 and a DFID score of 0.25, showcasing its potential for various applications.