Joint Face Detection and Facial Motion Retargeting for Multiple Faces

To produce a 3D virtual character's face expression of human’s natural face expressions, facial motion capture is the technique considered to be the most effective one, especially in terms of production speed. However, there are still some results showing that the expression is not so expressive, especially on the side of the 3D character which has a different facial features than the real models regarding to the application of it. In this research, the correction of the basic expressions of faces in the process of facial motion retargeting was done by using blendshape interpolation method that was based on fuzzy logic. Blendshape interpolation method is the method used to combine multiple shapes into one blend with the concept of interpolation. In this research, the process of blendshape meets the concept of linear interpolation which the movement of a point of vertexon blendshape used straight lines . Blendshape method will be run as a proofreader on the results of retargeting process. Theweighting of blendshape will be assigned automatically from the results of the calculation of fuzzy logic, which refers to the input of the marker position of the facial motion retargeting. This weight is then used to provide improvements to create more expressive expressions. This process will be easier and faster to do than doing customize one by one at the vertex point manually. To avoid the appearance of irregular motion (haphazard movement), it is necessary to give the limitation of the weight (weight constraint) with range of [0,1].Keywords : Blendshape, retargeting, fuzzy logic, facial motion capture.

Download Full-text

Lightweight and Real-Time Framework for Facial Motion Retargeting

2020 International Conference on Artificial Intelligence and Computer Engineering (ICAICE) ◽

10.1109/icaice51518.2020.00064 ◽

2020 ◽

Author(s):

Di Jiang ◽

Chengwen Zhang

Keyword(s):

Real Time ◽

Facial Motion ◽

Motion Retargeting

Download Full-text

Physiologic Studies Provide New Perspectives on Early Speech Development

Perspectives on Speech Science and Orofacial Disorders ◽

10.1044/ssod20.2.29 ◽

2010 ◽

Vol 20 (2) ◽

pp. 29-36

Author(s):

Erin M. Wilson ◽

Ignatius S. B. Nip

Keyword(s):

Motor Control ◽

Motor Development ◽

Motion Tracking ◽

Speech Development ◽

Speech Motor Control ◽

Therapeutic Approaches ◽

Facial Motion ◽

Disordered Speech ◽

Speech Motor ◽

Physiologic Data

Abstract Although certain speech development milestones are readily observable, the developmental course of speech motor control is largely unknown. However, recent advances in facial motion tracking systems have been used to investigate articulator movements in children and the findings from these studies are being used to further our understanding of the physiologic basis of typical and disordered speech development. Physiologic work has revealed that the emergence of speech is highly dependent on the lack of flexibility in the early oromotor system. It also has been determined that the progression of speech motor development is non-linear, a finding that has motivated researchers to investigate how variables such as oromotor control, cognition, and linguistic factors affect speech development in the form of catalysts and constraints. Physiologic data are also being used to determine if non-speech oromotor behaviors play a role in the development of speech. This improved understanding of the physiology underlying speech, as well as the factors influencing its progression, helps inform our understanding of speech motor control in children with disordered speech and provide a framework for theory-driven therapeutic approaches to treatment.

Download Full-text

Face Detection for AIBO Using RBF Network

IEEJ Transactions on Electronics Information and Systems ◽

10.1541/ieejeiss.130.2031 ◽

2010 ◽

Vol 130 (11) ◽

pp. 2031-2038

Author(s):

Kohki Abiko ◽

Hironobu Fukai ◽

Yasue Mitsukura ◽

Minoru Fukumi ◽

Masahiro Tanaka

Keyword(s):

Face Detection ◽

Rbf Network

Download Full-text

Speaker Identity Recognition by Acoustic and Visual Data Fusion through Personal Privacy for Smart Care and Service Applications

Journal of Imaging Science and Technology ◽

10.2352/j.imagingsci.technol.2020.64.4.040404 ◽

2020 ◽

Vol 64 (4) ◽

pp. 40404-1-40404-16

Author(s):

I.-J. Ding ◽

C.-M. Ruan

Keyword(s):

Face Detection ◽

Speaker Recognition ◽

Visual Information ◽

Classification Tree ◽

Gaussian Mixture ◽

Recognition Method ◽

Indoor Space ◽

Identity Recognition ◽

Visual Identity ◽

Speaker Classification

Abstract With rapid developments in techniques related to the internet of things, smart service applications such as voice-command-based speech recognition and smart care applications such as context-aware-based emotion recognition will gain much attention and potentially be a requirement in smart home or office environments. In such intelligence applications, identity recognition of the specific member in indoor spaces will be a crucial issue. In this study, a combined audio-visual identity recognition approach was developed. In this approach, visual information obtained from face detection was incorporated into acoustic Gaussian likelihood calculations for constructing speaker classification trees to significantly enhance the Gaussian mixture model (GMM)-based speaker recognition method. This study considered the privacy of the monitored person and reduced the degree of surveillance. Moreover, the popular Kinect sensor device containing a microphone array was adopted to obtain acoustic voice data from the person. The proposed audio-visual identity recognition approach deploys only two cameras in a specific indoor space for conveniently performing face detection and quickly determining the total number of people in the specific space. Such information pertaining to the number of people in the indoor space obtained using face detection was utilized to effectively regulate the accurate GMM speaker classification tree design. Two face-detection-regulated speaker classification tree schemes are presented for the GMM speaker recognition method in this study—the binary speaker classification tree (GMM-BT) and the non-binary speaker classification tree (GMM-NBT). The proposed GMM-BT and GMM-NBT methods achieve excellent identity recognition rates of 84.28% and 83%, respectively; both values are higher than the rate of the conventional GMM approach (80.5%). Moreover, as the extremely complex calculations of face recognition in general audio-visual speaker recognition tasks are not required, the proposed approach is rapid and efficient with only a slight increment of 0.051 s in the average recognition time.

Download Full-text