This repo serves as a storage location for the proposal, report, and data for final submission of the Capstone Project of the Udacity Machine Learning Engineer Nanodegree.
All data used in this project can be found in the data folder. If you try to reproduce the results in Capstone.ipynb, you will need to update the "real" images in the folders data/train, data/val, and data/test, replacing them with the corresponding data from the folders in real_round1. Other than that, you should be able to run the notebooks as-is. See the following section for a caveat to that.
Please note that the models were originally trained on CPU-based instances in Amazon Sagemaker. The SVM classifier was trained using a more powerful instance, so you may run out of memory depending on what machine you use to run that notebook.
The following Python packages were used in this project. Most were available in the conda_pytorch_py36 kernal available in Amazon SageMaker.
- Matplotlib
- NumPy
- PyTorch
- requests