Fingerprint recognition using CNN
Let's assume that left image is a new input from user, center and right images are stored in database.
Title of figures [122 0 0 1] means, left to right order:
- subject_id
- gender (male 0, female 1)
- left_or_right_hand (left 0, right 1)
- finger index (0-4)
Applied some augmentation(gaussian blur, zoom, translation, rotation..) to input(left image) for wild environment.
Center image is the answer so it has 99% confidence, on the other hand, right image is wrong so that has 0%.
- Python
- numpy
- keras
- matplotlib
- sklearn
- imgaug
Sokoto Coventry Fingerprint Dataset (SOCOFing) https://www.kaggle.com/ruizgara/socofing/home