Guess The Age 2021 is an international contest meant for teams able to propose methods based on modern Deep Convolutional Neural Networks (DCNNs) for age estimation from facial images. In order to allow the teams to train effective models, the Mivia Age Dataset, including 575.073 images annotated with age labels, was provided as training set; it is among the biggest publicly available datasets of faces in the world with age annotations. The performance of the methods submitted by the teams have been evaluated on a test set of more than 150.000 images, different from the ones available in the training set; a new index, called AAR, which takes into account the age estimation accuracy, namely the average error, and the regularity, i.e. the standard deviation of the error, has been adopted for drawing up the final ranking. The BTWG team, winner of the contest, achieved an impressive AAR equal to 7.94, with a novel method demonstrating an impressive accuracy and regularity in facial age estimation.

Guess the Age 2021: Age Estimation from Facial Images with Deep Convolutional Neural Networks

Greco A.
2021

Abstract

Guess The Age 2021 is an international contest meant for teams able to propose methods based on modern Deep Convolutional Neural Networks (DCNNs) for age estimation from facial images. In order to allow the teams to train effective models, the Mivia Age Dataset, including 575.073 images annotated with age labels, was provided as training set; it is among the biggest publicly available datasets of faces in the world with age annotations. The performance of the methods submitted by the teams have been evaluated on a test set of more than 150.000 images, different from the ones available in the training set; a new index, called AAR, which takes into account the age estimation accuracy, namely the average error, and the regularity, i.e. the standard deviation of the error, has been adopted for drawing up the final ranking. The BTWG team, winner of the contest, achieved an impressive AAR equal to 7.94, with a novel method demonstrating an impressive accuracy and regularity in facial age estimation.
978-3-030-89130-5
978-3-030-89131-2
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11386/4779986
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact