Probing of the dataset with new individuals
After adding new data into their respective folders of either probe or gallery images, we ran the RecognitionGUI software, allowing for predictions of individuals. The RecognitionGUI software provides scores for each individual in the gallery, assigning them a number based on their facial biometric similarities to the individual in the probe. Similarity scores are provided for the top five ranked matches; top matches are confirmed by visual check in the output graphic.
To add new individuals to the gallery database, a ‘match’ probe photo can be automatically merged to its matching file. If the new individual does not match previous individuals in the dataset, a novel ID is created, and the seal photos and ID are added to the existing database. To speed up this process, we created a quick Python program, Tkinter GUI .