SignatureDetection

this project is influence by Learning Features for Offline Handwritten Signature Verification with CNN paper.

in the paper they proposed a multi-task feature extractor.

we did Implementation in 3 different way:

1. Training Separated CNN as feature extractor, and then using SVMs for classification.
2. Training shared CNN as feature extractor, and then using SVMs for classification.
3. Training shared CNN as feature extractor, and then using RNNs (LSTM, GRU, ...) for classification.

the documentation and presentation for each part is included.

requirements to run:

• tensorflow 1.x
• keras
• opencv