sohankandagatla/Handwritten-Amharic-Character-Recognition

A Django web app that recognizes handwritten Amharic characters using a hybrid XGBoost + VGG16 model, with dataset import and trainable accuracy from the admin panel. Create virtual environment if not there by typing script: python -m venv amharic_venv Activate the env amharic_venv\Scripts\activate Then you create the database in mysql client as: create database amharic_recognition_db; Install requirements : pip install -r requirements.txt Run python manage.py makemigrations & python manage.py migrate Dataset Setup : If you already have the dataset, skip this step entirely. If you don't have the dataset, run the following commands in order: python scripts/download_real_dataset.py python scripts/import_real_amharic.py python manage.py collectstatic If you can't sign in with an existing ID, create a superuser: python manage.py createsuperuser Runserver: python manage.py runserver You can train the model from the admin panel. For best results, use XGBoost + VGG16 — it takes approximately 1 hour to train. The more you train it, the higher the accuracy you'll get. Login Credentials If a superuser already exists, log in with: Admin — Username: admin | Password: admin For a regular user account: User — Username: user | Password: Uer@123