This repo contains information about new research field of machine learning that is handwritten character recognition.
- Dataset Preparation: Preprocessed datasets of handwritten characters for training the machine learning model.
- Model Training: Code files for training a deep learning model using popular frameworks such as TensorFlow or PyTorch.
- Model Evaluation: Evaluation scripts to measure the model's performance and accuracy.
- User Interface Development: Code and resources for building a user-friendly interface for the app.
- Deployment and Usage: Guidelines and scripts for deploying the app on a server or cloud platform.
To get started with the Handwritten Character Recognition app, follow these steps:
-
Clone the repository:
git clone https://github.com/Tirth132108/Handwritten-Character-Recognitioin.git -
Dataset Preparation: Use the provided preprocessed datasets or replace them with your own data, ensuring proper annotation.
-
Model Training: Use the training code files to train a deep learning model on the dataset. Customize the model architecture and parameters as needed.
-
Model Evaluation: Run the evaluation scripts to assess the model's performance on separate validation datasets. Make necessary adjustments to improve accuracy.
-
User Interface Development: Utilize the code and resources to develop a user-friendly interface for the app. You can choose a suitable framework like Flask or Django.
-
Deployment and Usage: Follow the provided guidelines or scripts to deploy the app on a server or cloud platform. Ensure proper documentation to guide users on app usage.
To run the project files, put Handwritten Character Recognition.ipynb and model1.h5 in the same directory.
OpenCV-Python has to be installed in your system. If not that find the installation here https://pypi.org/project/opencv-python/
You also need Numpy, matplotlib, tensor flow and imutils library in your device.
After installing all of them just run the Handwritten Character Recognition.ipynb file.
I provided some example images for prediction, however if you want you can download any jpeg or png files and just change the path of the image in image_path = “image path” and you are good to go.
Hope you find my project interesting.