Spark_Image_Classification

Example Image Classification Program Using Spark

Requires Intel Analytics Zoo

Link to Intel Analytics Zoo GitHub Project

Link to Install Instructions for Intel Analytics Zoo

Requires Pyspark

Link to Apache Spark Website

Link to Pyspark Documentation

Requires Keras

Link to Keras GitHub

Link to Keras Documentation

Reads Files from

./caltech-256-image-dataset/256_ObjectCategories/*/*

Dataset Download

Link to Caltech256 Website for Dataset

Running the code

Create the docker container using

$ docker run -it -p 8888:8888 -e ACCEPT_EULA=yes microsoft/mmlspark

Add python and model files to docker container

$ docker cp [image-dataset-folder] [container id]:\notebooks\

$ docker cp [local-ipynb-file] [containerid]:\notebooks\

The .ipynb notebook can be started in docker container and should run without issue.

To run the .py file use docker exec:

$ docker exec [container id] spark-submit --packages Azure:mmlspark:0.13 /notebooks/WorksInDocker-ExistingModel.py