Chalearn Workshop on Face Spoofing Attack @CVPR19
Long Beach, CA
Face anti-spoofing detection is an crucial procedure in biometric face recognition systems. Previous competitions (i.e. ICB 2013 facial spoofing attacks, IJCB 2011 facial spoofing attacks, IJCB 2017 competition ) focused on 2D face spoofing attacks, and most published works focus on one single modality, such as rgb or depth face spoofing detection. However, one single modality may not capture rich enough face and environment information. Fortunately, according to the new development of camera sensors, face image with multi modalities are captured conveniently in a low cost. Therefore, We are organizing an academic competition and workshop focus on multi-modal (RGB+depth+IR) face anti-spoofing detection in videos. A new large-scale multimodal face spoofing attack datasets is released and used in the competition containing more than 2000 recorded subjects with multiple modalities.
We propose a challenge that aims at compiling the latest efforts and research advances from the computational intelligence community in creating fast and accurate face spoofing detection algorithms. The methods will be evaluated on a large, newly collected and annotated dataset.
The challenges is running in the CodaLab platform (https://competitions.codalab.org/competitions/20853), and results will be presented at the CVPR 2019 ChaLearn LAP associated workshop. Participants obtaining the best results will be invited to submit a paper to associated workshop and extended versions to a dedicated Special Issue in a top tier journal (TBA). There will be prizes and travel grants from BAIDU. We will provide cash, travel grants for top 3 winners (1st: 1000$+500$, 2nd:600$+500$, 3rd: 300$+500$). And the best workshop paper will be also rewarded in cash (500$+500$).
We also solicit submissions in all aspects of facial biometric systems and attacks. Most notably, the following are the main topics of interest:
- Novel methodologies on anti-spoofing detection in visual information systems.
- Studies on novel attacks to biometric systems, and solutions
- Deep learning methods for biometric authentication systems using visual information
- Novel datasets and evaluation protocols on spoofing prevention on visual and multimodal biometric systems
- Methods for deception detection from visual and multimodal information
- Face antispoof attacks dataset (3D face Mask, multimodal).
- Deep analysis reviews on face antispoofing attacks.
- Generative models (e.g. GAN) for spoofing attacks.
We are accepting submissions for revision up to 8 pages (same formatting instructions as CVPR). CMT submissions will be reviewed by 3 members of PC committee.