Unconstrained Ear Recognition using Combination of Deep Learning and Handcrafted Features

Tech ID: 17B171

Competitive Advantages

  •     Enables better identification
  •     Uses a plurality of landmarks for increased specificity
  •     Faster compared to similar methods
  •     Less computationally intensive

Summary

Our inventors have developed a biometric identification system to identify an ear image. This includes convolutional neural network (CNN) and handcrafted features that accurately identifies landmarks and discriminatory features. The ensures that even when the ears are flipped the system can still identify them. This system also combines a relatively fast process with a small amount of computational intensity. This enables a better security measure for devices such as mobile phones and laptops. This invention has immense application in the field of biometrics and ensures more data security. 

The Basic Process the Biometric System Uses to Identify a Specific Ear   

Desired Partnerships

  • License
  • Sponsored Research
  • Co-Development

Technology Transfer
TTOinfo@usf.edu
(813) 974-0994

Patents