Prerequisite: Tensorflow installed. Anaconda installed. CV2 installed. Numpy installed. Data set.
Process of working:
-
You will collect signatures from persons, let say you have collected signatures from two person AAA, BBB.
-
Create a folder named AAA and BBB.
-
Then put person AAA signatures to AAA folder and BBB signatures to BBB folder.
-
Then cut 20% of signatures photo from AAA and paste it in test_own_data folder randomly and also same amount from BBB.
-
Change the epoch size in last line of SystemTraining.py with your data suitability.
-
Then Run the file SystemTraining.py and leave it for enough time to run. It will generate some files of learning.
-
predictAll.py will predict all the signatures from test_own_data folder.