Hackwestx2024Juan-Tyler-Shelton

The hackwestx 2024 project for Juan, Tyler, and Shelton

Cowboy Forum Project 🤠

Our project consists of developing a fun and engaging forum for our fellow cowboys. When creating an account, users receive a "verified" tag, which can range between Space_Cowboy, Sheriff, and Raider. These tags are assigned based on a facial scan that classifies the user into one of these categories, adding a fun twist to the traditional profile creation process.

Technologies Used

The project used both MATLAB and Python for model generation. The facial classification model was trained on approximately 13,000 open-source images. Model accuracy reflects a result of 70%.

MATLAB

MATLAB was employed for the classification of the dataset into three distinct groups:

• Space_Cowboy
• Sheriff
• Raider

Criteria for Dividing Faces into the Three Groups

The division of faces was based on similarity in facial features, extracted using Histogram of Oriented Gradients (HOG) feature descriptors, followed by K-means clustering.

Grouping Process

• The facial features were extracted using HOG, and the faces were grouped into three categories by clustering their features using K-means.

• Faces with similar structural and textural patterns were grouped into one of the following three categories:

  • Space_Cowboy: Faces with certain angular features or specific textures.
  • Sheriff: A distinct group with different facial patterns compared to Space_Cowboy.
  • Raider: Faces grouped together based on another unique set of features.

The grouping was done algorithmically based on feature similarity rather than manual labeling.

Face Classification

How to Run the Face Model:

For this project, a Jupyter Notebook was used, and you will need the following libraries installed:

pip install opencv-python matplotlib face_recognition tensorflow scipy numpy Pillow h5py

Install/Run

• First download all files and unzip them into their own folder
• Install MongoDB using instructions from https://www.mongodb.com/docs/manual/tutorial/install-mongodb-on-windows/
• Run the following command bash mongod --dbpath <directiory_to_AppDb> where is replaced with the file path to the AppDb folder
• Open ./AppDbApi/AppDbApi.sln in Visual Studio 2022 with Asp.NET core installed and run the project
• Open ./App/App.sln in a separate instance of Visual Studio and run the project ## References Images used for face classification model: https://www.kaggle.com/datasets/jessicali9530/lfw-dataset

Built With

• asp.net
• c#
• matlab
• mongodb