transfer-learning-with-gluon

This project shows how to apply transfer learning with MXNet Gluon, on the husky or akita classification problem.

Prerequisite

1. install MXNet: pip install mxnet
2. install Jupyter Notebook: pip install jupyter

Project structure

1. data directory:
   1. img: image data to be used in generating rec file
   2. train, sample, test, validation: rec files for training
2. prepare-dataset.ipynb: preprocess and prepare datasets for training
3. train.ipynb: build model and define training process
4. generate_rec.sh: script to generate rec files
5. im2rec.py: script provided by MXNet tools

How to use this project

This project provides the simple guidance to apply transfer learning.

Users are expected to gather their own dataset and follow the notebooks.

For more details, please refer to this post