IJCB’s facial recognition adversarial attack challenge kicks off

This Monday saw the official kick-off of the 2025 Adversarial Attack Challenge (AAC), a competition aimed at strengthening biometric authentication systems against adversarial attacks.

The challenge is taking place as part of the International Joint Conference on Biometrics (IJCB) 2025 and is organized by Portugal-based authentication company Youverse and the Institute of Systems and Robotics at the University of Coimbra (ISR-UC) in Portugal. The competition opened on March 17th with the release of the dataset and Github instructions. Results are expected in June.

Adversarial attacks against facial recognition systems present a tough challenge as they can deceive deep neural networks used by these systems with subtle manipulations of images. The attacks can lead to identity fraud and unauthorized access.

“Adversarial attacks pose one of the most pressing challenges to modern biometric security,” explains Miguel Lourenço, Youverse’s chief product officer and head of AI. “The ability to detect and withstand adversarial manipulations is crucial to ensuring the reliability of biometric authentication.”

The 2025 Adversarial Attack Challenge (AAC) will focus on two key challenges: The Detection Track will allow participants to develop models capable of distinguishing between adversarially manipulated images and clean samples while the Resilience Track will ask competitors to create face recognition models that maintain high accuracy even when subjected to adversarial attacks.

The competition will grant monetary prizes of up to $3,500 for open-source solutions while the top three teams in each track will be invited to summarize their results in a peer-reviewed paper. The papers will be presented at the IJCB 2025, taking place September 8 to 11 in Osaka, Japan.

Aside from the competition, the initiative aims to establish a public adversarial attack library, open-source datasets and biometric security benchmarking tools, Youverse notes in a release.

Article Topics

adversarial attack  |  biometrics  |  biometrics research  |  facial recognition  |  IJCB  |  Youverse