So as I get more into biometrics I have to think of cool paper titles.  This is a cool work on iris recognition that uses the triplet loss to train the underlying neural network.  We managed to think of the name ThirdEye which is just about perfect.  Iris recognition is usually four steps:

1. Segmentation: split pixels into iris/non-iris,
2. Normalization: split the variable size image into a fixed dimension,
3. Feature extraction: remove noise and further reduce dimension, and
4. Comparison: compare the resulting template with a previous version.

Using the triplet loss with modern neural networks we challenge the belief that normalization is helpful (at least in all situations).  Source is here: https://github.com/sohaib50k/ThirdEye—Iris-recognition-using-triplets

 

We were inspired by a similar paper called “Influence of Segmentation on deep iris recognition performance” by Lozej, Stepec, Struc, and Peer that asked if segmentation was useful.