Skip to main content

Handwritten Digit Recognizer

Xircuits Handwritten Digit Classifier Project Template

This template allows you to train a handwritten digit classifier, using the Mnist Handwritten Digit Classification Dataset.

  • Dataset preparation: this section handles the dataset used in this template through multiple components.

    • DownloadDataset : download mnist dataset.
    • VisualizeData : visualize the training data (optional)

  • Model training: build and compile the model for training.

    • CreateModel : building a simple convolutional model.
    • TrainModel : training the model with training dataset.
    • PlotTrainingMetrics : evaluate training performance, by plotting the training loss and accuracy against the number of training epochs.
    • EvaluateModel : determine the model loss and accuracy based on the testing dataset.
    • SaveModel : save model in keras or tensorflow format.
    • ConvertTFModelToOnnx : convert TF model to onnx model to be used in other platforms.

Prerequisites​

You will need Python 3.9+.

Installation​

  1. Clone this repository
  2. Create virtual environments and install the required python packages.
pip install -r requirements.txt
  1. Run xircuits from the root directory
xircuits

Workflow in this Template​

mnist_classifier_template.xircuits​

  • In this template we used the mnist dataset from Tensorflow and perform a simple classification. You can further fine tune the model by modifying the hyperparameters.

Template