Feature Generation and Hypothesis Verification for Reliable Face Anti-spoofing

Authors: Shice Liu, Shitao Lu, Hongyi Xu, Jing Yang, Shouhong Ding, Lizhuang Ma1782-1791

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show our framework achieves promising results and outperforms the state-of-the-art approaches on extensive public datasets.
Researcher Affiliation Collaboration Shice Liu1*, Shitao Lu1,2*, Hongyi Xu1,3, Jing Yang1 , Shouhong Ding1 , Lizhuang Ma2,3 1Youtu Lab, Tencent, Shanghai, China 2East China Normal University, Shanghai, China 3Shanghai Jiao Tong University, Shanghai, China
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Our source code is available at https://github.com/lustoo/FGHV.
Open Datasets Yes Above all, we conduct the cross-dataset testing on four public datasets, i.e., OULU-NPU (denoted as O) (Boulkenafet et al. 2017), CASIA-MFSD (denoted as C) (Zhang et al. 2012), Idiap Replay-Attack (denoted as I) (Chingovska et al. 2012) and MSU-MFSD (denoted as M) (Wen et al. 2015). After that, the cross-type testing is carried out on the rich-type dataset, i.e., Si W-M (Liu et al. 2019).
Dataset Splits No The paper states 'select one dataset for testing and the other three datasets for training' for cross-dataset testing and 'select out one attack type as the unknown testing type and treat the others as the known training types' for cross-type testing, but it does not explicitly provide details about a separate validation split or how it's derived.
Hardware Specification Yes All experiments are conducted via Py Torch on a 32GB Tesla-V100 GPU.
Software Dependencies No The paper mentions using 'Py Torch' but does not specify its version number or versions of other software dependencies.
Experiment Setup Yes During the training period, the framework is trained with SGD optimizer where the momentum is 0.9 and the weight decay is 5e-4. The learning rate is initially 1e-3 and drops to 1e-4 after 50 epochs. The hyper-parameters λ1 and λ2 are both set to 1.