x1nzhe/numberrecognizer

NEU INFO5100 Final Project

Digit Recognizer with TensorFlow and JavaFX NEU INFO5100 Final Project

This project demonstrates a digit recognizer implemented using TensorFlow in Python for training and exporting the model, and Java with JavaFX for creating a user interface to draw digits and receive predictions.

Overview The project consists of two main parts:

Training the Model in Python: Utilized the TensorFlow library to create and train a convolutional neural network (CNN) on the MNIST dataset. The trained model is then saved in TensorFlow format for later use.

JavaFX User Interface: Developed a Java application with JavaFX to create a user-friendly interface for drawing digits. Integrated TensorFlow Java API to load the pre-trained model and obtain predictions. The application allows users to draw a digit on a canvas, and upon submission, the model predicts the digit.

Project Structure |-- java |-- src |-- main |-- java |-- com.neu.info5100.numberrecognizer |-- Launcher.java
|-- NumberRecognizerApplication |-- NumberRecognizerController.java |-- NumberRecognizerController |-- NumberRecognizerModel.java |-- TensorFlowModelLoader.java

            |-- com.neu.info5100.numberrecognizer
                |--NumberRecognizer-view-fxml
    |--Test Screenshots
|-- README.md                    

Run the Application Open the project in your preferred Java IDE. Build and run the Launcher.java file in src/main/java/com.neu.info5100.numberrecognizer The JavaFX application window will appear, allowing you to draw digits.