This folder contains the code to build a deep learning model on MNIST handwritten digit dataset and converting the trained model to TF Lite format.

• https://www.tensorflow.org/guide/datasets

• This code is checked on using native Python 3 with anaconda
• Create a conda virtual environment and install relevant packages using requirements.txt file

pip install requirements.txt

To run the code just execute

python main.py

On CPU the code might take few mins to run. However, if you use GPUs it should be much faster

Dataset used for this dataset is the standard MNIST handwritten digits dataset available in Tensorflow Datasets.

Code is pretty self explanatory. There are mainly 3 files in implementation:

• main.py -- Implements the main function and also implements model building and training routines.
• parameters.py -- Defines the parameters used in the code.
• utils.py -- Contains the helper functions for the code

This code implements graph freezing and optimizing but will have to use Tensorflow Optimization Converter Tool (toco) to convert the optimized graph to .tflite format.