This repository contains the final project submission of the group Average.Pandas for the course EEL5840 - Fundamentals of Machine Learning. In this project, we have developed a deep learning pipeline to classify images of 10 handwritten symbols.

This milestone is to be delivered at the end of the semester, Friday April 22 @ 11:59 PM. Find the complete rubric in the Canvas assignemtn.

The training data set is the same for every team in this course.

You can download the data in our Canvas page:

• "data_train.npy"
• "labels_train.npy"

Clone the repository

git clone https://github.com/EEL5840-EEE4773-Spring2022/final-project-code-and-report-average-pandas.git

Here is the list of packages that need to be installed to run this project.

• tensorflow-gpu

  conda install -c anaconda tensorflow-gpu

• keras

  conda install -c conda-forge keras

• opencv

  conda install -c conda-forge opencv

• scikit-learn

  conda install -c anaconda scikit-learn

• matplotlib

  conda install -c conda-forge matplotlib

• numpy

  conda install -c anaconda numpy

• scipy

  conda install -c anaconda scipy

Alternatively, you can create the python working environment from environment.txt file. However, since the original environment was created in a linux red hat machine, there are some dependencies on the operating system. Some of the packages might be missing in a windows machine when creating the environment.

1. Setup the environment

conda create --name average_pandas --file environment.txt

2. Activate the environment

conda activate average_pandas

This repository contains the follwoing files:

• "train.py": training script
• "test.py": testing script
• "modelCNN.hdf5": pretrained model
• "Data" directory: contains example test data and labels

The train.py file is used to train the dataset with our developed deep convolutional neural network. The directory containing the training data and labels need to be modified in the variable file_path. The trained model is saved in modelCNN.hdf5 file.

The model can be tested using the test.py file. Similar to the training phase, the directory containing the test data and labels need to be modified in the variable file_path. We provided a sample test dataset in my_test_images.npy and my_test_labels.npy files for testing our model. Running the test.py script will provide the accuracy, predicted and true labels as output.

Your Name - Hasan Al-Shaikh ([email protected]), Shuvagata Saha ([email protected]), Md Kawser Bepary ([email protected])

Project Link: https://github.com/EEL5840-EEE4773-Spring2022/final-project-code-and-report-average-pandas

We would like to thank the course instructor, Dr. Catia Silva, and the teaching assistant, Haotian Yue, for their continuous support and motivation to finish the project on time. We would like to also thank our friend Dipayan Saha for his suggestions and feedback in various stages of the project.