The Project is conducted using a benchmarking Signature Verification Dataset:

• The aim of the project is to evaluate the performance of some of the prominent available Deep Convolutional Neural Networks in Writer Independent Signature Verification/Recognition for automatic recognition of the signatures. This is done using four Pre-trained Models:

1. MobileNetV2

2. InceptionV3

3. ResNet50

1. MobileNetV2 Convolution Neural Network

• Accuracy = 82.20 %
• 

2. InceptionV3 Convolution Neural Network

• Accuracy = 73.60 %
• 

3. ResNet50 Convolution Neural Network

• Accuracy = 85.20 %
•