27 February 2023 in IEEE Transactions on Information Forensics and Security
DeepFake detection aims to differentiate falsified faces from real ones. Most approaches formulate it as a binary classification problem by solely mining the local artifacts and inconsistencies of face forgery, which neglect the relation across local regions. Although several recent works explore local relation learning for DeepFake detection, they overlook the propagation of relational information and lead to limited performance gains. To address these issues, this paper provides a new perspective by formulating DeepFake detection as a graph classification problem, in which each facial region corresponds to a vertex. But relational information with large redundancy hinders the expressiveness of graphs. Inspired by the success of masked modeling, we propose Masked Relation Learning which decreases the redundancy to learn informative relational features. Specifically, a spatiotemporal attention module is exploited to learn the attention features of multiple facial regions. A relation learning module masks partial correlations between regions to reduce redundancy and then propagates the relational information across regions to capture the irregularity from a global view of the graph. We empirically discover that a moderate masking rate (e.g., 50%) brings the best performance gain. Experiments verify the effectiveness of Masked Relation Learning and demonstrate that our approach outperforms the state of the art by 2% AUC on the cross-dataset DeepFake video detection. Code will be available at https://github.com/zimyang/MaskRelation.
17 March 2022 in IEEE Transactions on Information Forensics and Security
Heterogeneous Face Recognition (HFR) aims to match faces across different domains ( e.g. , visible to near-infrared images), which has been widely applied in authentication and forensics scenarios. However, HFR is a challenging problem because of the large cross-domain discrepancy, limited heterogeneous data pairs, and large variation of facial attributes. To address these challenges, we propose a new HFR method from the perspective of heterogeneous data augmentation, named Face Synthesis with Identity-Attribute Disentanglement (FSIAD). Firstly, the identity-attribute disentanglement (IAD) decouples face images into identity-related representations and identity-unrelated representations (called attributes), and then decreases the correlation between identities and attributes. Secondly, we devise a face synthesis module (FSM) to generate a large number of images with stochastic combinations of disentangled identities and attributes for enriching the attribute diversity of synthetic images. Both the original images and the synthetic ones are utilized to train the HFR network for tackling the challenges and improving the performance of HFR. Extensive experiments on five HFR databases validate that FSIAD obtains superior performance than previous HFR approaches. Particularly, FSIAD obtains 4.8% improvement over state of the art in terms of VR@FAR=0.01% on LAMP-HQ, the largest HFR database so far.
12 June 2020 in European Journal of Medicinal Plants
Aims: The ripe fruit of Siraitia grosvenorii (Swingle) C. Jeffrey (Cucurbitaceae) is used in Traditional Chinese Medicine (TCM) for treatment of diabetes. The ripe fruit is very edible, nutritious and of high sweet taste, while the unripe fruit which is usually very bitter and not edible, has not been widely studied. Therefore, the aim of this study is to evaluate the antidiabetic activity of the unripe fruit of S. grosvenorii and to compare it with that of the ripe fruit as well as to determine their possible mechanism of action. Methodology: The method of streptozotocin-nicotinamide induced diabetic animal model in mice was employed and mice with blood glucose greater than 9.0 mmol/l were selected for treatment with unripe fruit extract (UFE) and ripe fruit extract (RFE) for 24 days. Alpha-glucosidase and alpha amylase enzymes inhibitory assays for determination of possible mechanism of action, were performed. Results: The RFE exhibited better efficacy in antidiabetic activity than the UFE, though the later displayed better potency in activity. The RFE, UFE (300 mg/kg) and glibenclamide exhibited significant (P<0.05) percentage blood glucose reduction (PBGR) of 67.04, 60.69 and 62.88%, respectively, in diabetic mice. Alpha glucosidase assay showed median inhibitory concentration (IC50) value of 7.60 and 6.62 mg/ml for RFE and UFE, respectively, while the alpha amylase inhibitory assay exhibited percentage inhibition of 84.78 and 95.10% for RFE and UFE (25 mg/ml), respectively. Conclusion: The ripe and unripe extracts of S. grosvenorii exhibited antidiabetic activities, and RFE proved a better efficacious activity than UFE, which showed more potent activity. In addition, the antidiabetic activity is likely being mediated through alpha glucosidase and alpha amylase enzymes inhibition.