We are now accepting submissions for the DFW2019 competition! Deadline: July 20, 2019!
With recent advancements in deep learning, the capabilities of automatic face recognition has been significantly increased. However, face recognition in unconstrained environment with non-cooperative users is still a research challenge, pertinent for users such as law enforcement agencies. While several covariates such as pose, expression, illumination, aging, and low resolution have received significant attention, “disguise” is still considered an arduous covariate of face recognition. Disguise as a covariate involves both intentional and unintentional changes on a face through which one can either obfuscate his/her identity or impersonate someone else’s identity. The problem can be further exacerbated due to unconstrained environment or “in the wild” scenarios. However, disguise in the wild has not been studied in a comprehensive way, primarily due to unavailability of such as database. As part of the International Workshop on Disguised Face in the Wild at ICCV2019, a competition is being held in which participants are asked to show their results on the Disguised Faces in the Wild (DFW) 2019 database.
Researchers who submit to the competition are highly encouraged to submit their paper to the DFW Workshop@ICCV2019 as well! The DFW 2019 dataset, protocols, and instructions for the competition will be released soon. Please fill this form to get all the updates!
The DFW2019 competition builds upon the DFW2018 competition, and encourages researchers to develop algorithms robust to disguise variations. To this effect, we have prepared a novel DFW2019 dataset which will form the test set for this competition. The training partition of the DFW2018 dataset can be used as the training set for this competition, while test set of the DFW2018 dataset will form the validation set for this competition. That is, for the DFW2019 competition:
The competition is now live! Please register below to participate and get details regarding obtaining the datasets:
The DFW2019 dataset contains over 3800 images of 600 subjects, encompassing different disguise variations including variations due to bridal make-up and plastic surgery. The dataset contains anonymized images with names 0.jpg to 3839.jpg. For this competition, the submitted algorithms will be evaluated on the DFW2019 dataset. The dataset is available as a password protected zip file along with the other files. The password is available to the participant after filling the registration form.
The DFW2018 dataset contains over 11,000 images of 1000 subjects. The dataset follows a pre-defined protocol, where images of 400 subjects form the training set, and the remaining 600 subjects constitute the test set. For this competition, the training set corresponds to the training partition of the DFW2018 dataset and the validation set corresponds to the test set of the DFW2018 dataset.
The dataset contains a folder for each subject, which consists of the subject normal, validation, disguised, and impersonator images. The nomenclature of the dataset is as follows:
Participants are encouraged to refer to the following two papers for the DFW2018 dataset:
The dataset is available as a password protected zip file along with the other files. The password is available to the participant after filling the registration form.
At the time of submission, participants will be required to submit the following:
Evaluations will be performed on the DFW2019 dataset for the following four protocols: