This repository contains the code and instructions to use the trained CNN models described in [1] to extract features for Offline Handwritten Signatures. It also includes links to download extracted features from the GPDS, MCYT and CEDAR datasets.
Based on research paper by-
[1] Hafemann, Luiz G., Robert Sabourin, and Luiz S. Oliveira. "Learning Features for Offline Handwritten Signature Verification using Deep Convolutional Neural Networks" http://dx.doi.org/10.1016/j.patcog.2017.05.012 (preprint)
Topics:
- Installation: How to set-up the dependencies / download the models to extract features from new signatures
- Usage: How to use this code as a feature extractor for signature images
- Datasets: Download extracted features (using the proposed models) for the GPDS, MCYT and CEDAR datasets (.mat files - do not require any pre-processing code)
Follow these steps to install the system.
The following libraries are required
- Scipy version 0.18
- Pillow version 3.0.0
- OpenCV
- Theano
- Lasagne
They can be installed by running the following commands:
pip install opencv-python
pip install scipy
pip install pillow
pip install "Theano==0.9"
pip install https://github.com/Lasagne/Lasagne/archive/master.zip
pip install django
pip install requests
We tested the code in Mac OSx High Serria. This code can be used with or without GPUs - to use a GPU with Theano, follow the instructions in this link. Note that Theano takes time to compile the model, so it is much faster to instantiate the model once and run forward propagation for many images (instead of calling many times a script that instantiates the model and run forward propagation for a single image).
- Clone (or download) this repository
- Download the pre-trained models from the project page
- Save / unzip the models in the "models" folder
Or simply run the following:
cd sbi/app_backend/models
wget "https://storage.googleapis.com/luizgh-datasets/models/signet_models.zip"
unzip signet_models.zip
The following code (from example.py) shows how to load, pre-process a signature, and extract features using one of the learned models:
git clone https://github.com/adarshsrivastava11/sbi.git
cd sbi
cd app_backend
python manage.py runserver
To facilitate further research, we are also making available the features extracted for each of the four datasets used in this work (GPDS, MCYT, CEDAR, Brazilian PUC-PR), using the models SigNet and SigNet-F (with lambda=0.95).
| Dataset | SigNet | SigNet-F |
|---|---|---|
| GPDS | GPDS_signet | GPDS_signet_f |
| MCYT | MCYT_signet | MCYT_signet_f |
| CEDAR | CEDAR_signet | CEDAR_signet_f |
There are two files for each user: real_X.mat and forg_X.mat. The first contains a matrix of size N x 2048, containing the feature vectors of N genuine signatures from that user. The second contains a matrix of size M x 2048, containing the feature vectors of each of the M skilled forgeries made targeting the user.
The code would not have been possible without these works:
[1] Hafemann, Luiz G., Robert Sabourin, and Luiz S. Oliveira. "Learning Features for Offline Handwritten Signature Verification using Deep Convolutional Neural Networks" http://dx.doi.org/10.1016/j.patcog.2017.05.012 (preprint)
The following research were responsible for generating the dataset:
GPDS: Vargas, J.F., M.A. Ferrer, C.M. Travieso, and J.B. Alonso. 2007. “Off-Line Handwritten Signature GPDS-960 Corpus.” In Document Analysis and Recognition, 9th I nternational Conference on, 2:764–68. doi:10.1109/ICDAR.2007.4377018.
MCYT: Ortega-Garcia, Javier, J. Fierrez-Aguilar, D. Simon, J. Gonzalez, M. Faundez-Zanuy, V. Espinosa, A. Satue, et al. 2003. “MCYT Baseline Corpus: A Bimodal Biometric Database.” IEE Proceedings-Vision, Image and Signal Processing 150 (6): 395–401.
CEDAR: Kalera, Meenakshi K., Sargur Srihari, and Aihua Xu. 2004. “Offline Signature Verification and Identification Using Distance Statistics.” International Journal of Pattern Recognition and Artificial Intelligence 18 (7): 1339–60. doi:10.1142/S0218001404003630.
The source code is released under the BSD 2-clause license. Note that the trained models used the GPDS dataset for training (which is restricted for non-comercial use).