Deep Learning solution built in PyTorch

The project is a deep learning solution for the problem of classifying different types of ID-Cards like Pakistani CNIC, Passport etc. Specific fields on the ID-Card can also be segmented out, like Name, DoB, etc. ResNet18 model is trained on custom dataset for the classification task. Mask RCNN model is also trained on same custome dataset for the segmentation task.

• Python
• PyTorch

A Jupyter Notebook has been provided will all the required steps.

• Google Colab or Jupyter Notebook

The installation steps are provided in the Jupyter Notebook.

1. Open up the notebook in Google Colab using your google account.

2. Place the custom dataset and annotations in your google drive.

3. Mount your drive and execute the notebook cells to train and test the models.

   from google.colab import drive
   drive.mount('/content/drive')

Distributed under the MIT License.

You can view the Jupyter Notebook here.