FacEDiM: A Face Embedding Distribution Model for Few-Shot Biometric Authentication of Cattle
This addresses biometric authentication for livestock management, but it is incremental as it applies an existing method to a new domain.
The paper tackles few-shot biometric authentication for cattle by using Mahalanobis distance with pre-trained CNN embeddings, achieving a FRR of 1.25% at a FAR of 1.18% on a dataset of 20 cattle identities.
This work proposes to solve the problem of few-shot biometric authentication by computing the Mahalanobis distance between testing embeddings and a multivariate Gaussian distribution of training embeddings obtained using pre-trained CNNs. Experimental results show that models pre-trained on the ImageNet dataset significantly outperform models pre-trained on human faces. With a VGG16 model, we obtain a FRR of 1.25% for a FAR of 1.18% on a dataset of 20 cattle identities.