… Because scarce samples insufficiently represent potential variations, the augmentation …
unconstrained palmprint databases. We aim to address FS-UPR via contrastive augmentation …

… Flipping and rotating are not suitable for the data augmentation in palmprint recognition,
and the main reason lies in that they destroy the orientation information of different textures, …

… augmentation method for palmprint recognition. Although DCGAN has been successfully used
to generate samples for training data augmentation, it … with low contrast palmprint images. …

… a novel Contrastive Learning-based Finger-Vein image recognition with Automatic Adversarial
Augmentation, namely CL3A-FV. Instead of the predefined data augmentation operations…

… to verify personal identify include face and fingerprint recognition that have been widely
applied in … 2) To perform palm-vein data augmentation, we explore a GAN to learn the mask …

… From the results, for Blue, the optimal accuracy is 100% on 5-way, 5-shot recognition using
similarity loss and contrastive loss. For Green, the optimal accuracy is 100% both on 5-way, 3-…

… contrast, brightness, sharpness or noises. Therefore, for the sake of enhancing the robustness
of learned CNN, we conducted the following data augmentation [… adopted contrastive loss …

… , fingerprint samples are easy to copy from the fingerprint residues on a sensor. By contrast,
… , we perform image augmentation using operators such as contrast, brightness, posterize, …

… learning process by performing data augmentation. In particular, each training sample (…
images across different datasets have different contrast, brightness, sharpness, quality of …

… see that the contrast between the palmprint and palm vein … recognition. We developed
a novel wavelet-based image fusion method where line-preservation and contrast-enhancement …