Age and gender-based human face reconstruction from single frontal image

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216224%3A14330%2F20%3A00115015" target="_blank" >RIV/00216224:14330/20:00115015 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/article/10.1007/s11042-018-6869-5" target="_blank" >https://link.springer.com/article/10.1007/s11042-018-6869-5</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/s11042-018-6869-5" target="_blank" >10.1007/s11042-018-6869-5</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Age and gender-based human face reconstruction from single frontal image

Popis výsledku v původním jazyce

We present an approach for the human face reconstruction from a single frontal image for the use in forensic anthropology when the subject’s age and gender is known. In our approach we build a database of several depth images per each age and gender group pair, marked with facial landmarks. To reconstruct a 3D facial model from an unknown frontal image we search the most similar face in the depth database based on the automatically detected landmarks and assign its depth to the model. In the evaluation part, we compared our approach to a recent automatic convolutional neural network based algorithm and a semi-automatic approach, where landmarks are required to be detected manually. In contrast to other tested approaches our algorithm can estimate all major components, such as eyes, nose and mouth, evenly. Thanks to the external depth database, it can also reconstruct human faces from images with partial facial occlusions and uneven lighting. Additionally, we have found that a single depth image provides a good approximation of the human face and a combination of multiple precomputed depth images has a little impact on the final 3D face reconstruction result. Speed measurements show that our algorithm provides a quick and a fully automatic way to reconstruct a human face from a single frontal image for the use in forensic anthropology.

Název v anglickém jazyce

Age and gender-based human face reconstruction from single frontal image

Popis výsledku anglicky

We present an approach for the human face reconstruction from a single frontal image for the use in forensic anthropology when the subject’s age and gender is known. In our approach we build a database of several depth images per each age and gender group pair, marked with facial landmarks. To reconstruct a 3D facial model from an unknown frontal image we search the most similar face in the depth database based on the automatically detected landmarks and assign its depth to the model. In the evaluation part, we compared our approach to a recent automatic convolutional neural network based algorithm and a semi-automatic approach, where landmarks are required to be detected manually. In contrast to other tested approaches our algorithm can estimate all major components, such as eyes, nose and mouth, evenly. Thanks to the external depth database, it can also reconstruct human faces from images with partial facial occlusions and uneven lighting. Additionally, we have found that a single depth image provides a good approximation of the human face and a combination of multiple precomputed depth images has a little impact on the final 3D face reconstruction result. Speed measurements show that our algorithm provides a quick and a fully automatic way to reconstruct a human face from a single frontal image for the use in forensic anthropology.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10200 - Computer and information sciences

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2020

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Multimedia Tools and Applications

ISSN

1380-7501

e-ISSN

1573-7721

Svazek periodika

79

Číslo periodika v rámci svazku

5-6

Stát vydavatele periodika

NL - Nizozemsko

Počet stran výsledku

26

Strana od-do

3217-3242

Kód UT WoS článku

000519410700009

EID výsledku v databázi Scopus

2-s2.0-85056907519