SigRec

TensorFlow Signature Recognition Code Documentation

This documentation explains signature recognition using the TensorFlow framework. The code employs a convolutional neural network (CNN) to classify signatures as genuine or forged.

Data Used: https://cedar.buffalo.edu/NIJ/data/

Model Trained: https://drive.google.com/file/d/1Qauvbk6vq-9oHqKAMfX1JgcOC6_uoVdm/view?usp=sharing

TensorFlow Signature Recognition Code Documentation

This documentation explains a Python script for signature recognition using the TensorFlow framework. The code employs a convolutional neural network (CNN) to classify signatures as genuine or forged.

Table of Contents

1. Introduction
2. Dependencies
3. GPU Configuration
4. Data Preprocessing
5. Model Setup
6. Model Training
7. Model Testing
8. Model Saving

1. Introduction

This code is designed to perform signature recognition using a machine learning model. The model is trained on a dataset of genuine and forged signatures and later tested for accuracy.

2. Dependencies

• TensorFlow: The deep learning framework used for building and training the model.
• NumPy: A library for numerical operations and array manipulations.
• Matplotlib: A visualization library for creating plots and charts.

The code imports necessary modules and libraries at the beginning.

3. GPU Configuration

This section configures GPU memory growth for TensorFlow. It ensures that GPU memory is allocated on an as-needed basis, which can help prevent memory overflow.

4. Data Preprocessing

The dataset is loaded using tf.keras.utils.image_dataset_from_directory. It assumes that the dataset is organized in directories with two subdirectories, one for genuine signatures and one for forged signatures. The dataset is split into training, validation, and test sets.

5. Model Setup

A convolutional neural network (CNN) model is set up using TensorFlow's Keras API. It consists of convolutional layers, max-pooling layers, flattening, and fully connected layers. The model is compiled with the Adam optimizer and binary cross-entropy loss.

6. Model Training

The model is trained on the training dataset using model.fit. Training data is specified by the train dataset, and validation data is specified by the val dataset. Training is performed for a specified number of epochs.

7. Model Testing

The trained model is evaluated on a test dataset. Precision, recall, and binary accuracy metrics are calculated using TensorFlow's built-in metrics classes (Precision, Recall, BinaryAccuracy). These metrics are updated for each batch in the test dataset, and the results are printed.

8. Model Saving

The trained model is saved to a file in the "models" directory using model.save. This saved model can be loaded later for inference or further training.

Note

• The code assumes that you have a dataset organized in the "data" directory with subdirectories for genuine and forged signatures.
• The model architecture and hyperparameters (e.g., number of layers, epochs) can be adjusted to improve performance.
• This is a simplified example for educational purposes. In a real-world application, more data preprocessing, data augmentation, and hyperparameter tuning may be necessary for better results.