This notebook is the implementation of Machine Learning model to classify writers based on their handwritings . Dataset used is IAM top 50 Dataset (where only 50 handwriting styles are considered).Here is the Link for the dataset .
I have built this model using Keras and Tensorflow . It is language independent as we are not considering letters in the classification , but patches of image 113X113 pixels are extracted from data and fit into the model for learning .
Optimizer : Adam
Convolution Layers : 3
Pooling: Max Pooling
Activation Functions Used : relu , softmax
For more details regarding model structure please refer to the this file .
This model gives 81% accuracy on test data for 9 epochs.