PalmNet: A CNN Transfer Learning Approach for Recognition of Young Children Using Contactless Palmprints

Recognition of children especially infants has received more attention nowadays. These are used in many applications such as law enforcement, health medical services, and vaccination tracking. Palmprint recognition has many unique advantages compared to other biometrics like richness of features, high userfriendliness, suitability for private security, etc. Acquiring the palmprints in a contactless way is very important in the context of pandemic situation even otherwise to avoid infection. In this paper, we created a dataset of palmprints of children using a mobile phone and proposed an approach for palmprint recognition using CNN transfer learning. Our approach is tested with IITD dataset having adult palmprints as well as compared with a state-of-the-art approach that uses infant palmprints. It is verified that our approach exhibit a good performance with a matching accuracy of 96% and Equal Error Rate (EER) of 0.02%.

Automated summary

Recognition of children's palmprints has gained attention, and the use of CNN transfer learning for palmprint recognition is proposed in this paper. The results demonstrate the effectiveness of the approach with a high matching accuracy of 96% and a low Equal Error Rate (EER) of 0.02%.