Run PyCon2016 Digit Recognition project using Docker
Summary:
In this era as we all know people are crazy about Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) terms so do I, so I thought to work on Machine Learning project. I started looking on many Blogs, Github repositories and found a very good project to start off and that was “PyCon2016 Digit Recognition” project originally build by Aleksei Tiulpin. When I was trying to run and learn this project, even if the project is very easy to understand but still I came across many challenges with configurations, installation of dependencies. So I tried to run the project on Docker so that it will be easy for others to test and learn. Throughout the blog, I’ll try to explain whatever I’ve done so far.
Prerequisite:
1. Anaconda-Navigator installed with Python3
2. Docker installed
Please follow the below steps to run the project and test:
1. Download this repository (DigitRecognition) in your local machine.
2. Launch Jupyter Notebook which opens the Jupyter notebook homepage in your default browser.
3. Navigate to the MNIST.ipynb file which is under PyCon2016 folder.
4. Open MNIST.ipynb into Jupyter notebook and run each cell one by one to build the predictive model, which will take some time. (If cell shows [ * ] it means the cell is still running)
5. Once all cells executed start the docker in your machine. ( Note: If you don’t have docker please install docker first).
6. Goto the terminal, navigate to the DigitRecognition folder which you have downloaded in Step 1 and type the below command. This will produce the docker image with all the necessary setup with the name mycon. (Note: Don’t
forget the dot(.) at the end).
docker build -t mycon .
7. Now once the docker image is ready. Run the docker image with the below command.
docker run -d -p 5000:5000 mycon
8. Now go to the browser and type the given URL: http://localhost:5000/
9. You’ll see something like this below if everything works :)
Note: I really appreciate Aleksei Tiulpin work for creating PyCon2016 repository here.
