Skip to content

README.md

Latest commit

 

History

History
45 lines (27 loc) · 1.21 KB

README.md

File metadata and controls

45 lines (27 loc) · 1.21 KB
Raw

Offline Signature Verification using Siamese Convolutional Neural Networks in Pytorch

Steps to produce the results

  1. Git clone the repository
  2. Run SignatureVerificatin.py file

Loss

Loss.

Accuracy

Accuracy

Architecture

Siamese convolutional Neural Network is used.

Dataset provided : Pair of images ( genuine- genuine / genuine - forge)

Custom Dataloader using Dataloader in pytorch.

Above results are for

Iteration : 500

Batch Size : 180

Resized image to 150 * 150 because of Processing time and memory limitations

English signature datasets.

System

These experiments were ran on 24GB RAM and Core i5 8th gen with Nvidia.

List of experiments you can perform using this code.

Increase the dataset size and Iterations.

Check for other accuracy measures to appropriately determine the model consistency.

Try running for different language datasets.

References

Siamese convolutional Neural Networks

SigNet

Dimensionality Reduction by learning an invariant mapping