Dataset-Creator

Application for creating numeric datasets from Images using holistic model of Mediapipe by saving landmark details in records. This WebApp can be used by ML researchers who are working majorly on facial landmarks and want to generate simple csv file directly.

Project Information:

Target Audience:

1. Deep Learning practioners and researchers working with Image Processing.
2. Data Preparation Teams struggling with available tools for classification problems.

Purpose: To improve data preparation process by facilitating users to create numeric records for their image dataset, easing the implementation of machine learning pipelines.

Check it out

Project Timeline:

1. Create Streamlit app and Image to landmarks to CSV Record.
2. Add Live Image Capture feature and make record addition easy.
3. Add Live Video Capture feature and add records from it. (Under Construction)

Instructions to run

• Pre-requisites:

  • The user must be familiar to data preparation methods and file formats.
  • Must have an IDE Installed(VSCode or PyCharm)
  • Basic knowledge of commands for using terminal

• Directions to setup/install

>> virtualenv venv
>> venv/scripts/activate

• Directions to execute

>> streamlit run main.py

Contributors

• KUSHAGRA SRIVASTAVA

Steps for beginners/non-ml students

1. Download the project folder/Clone the Repository.
2. Open Terminal and run venv/scripts/activate.
3. (venv) will appear before the directory.
4. Run streamlit run main.py.
5. Create you own CSV with the app.
6. Mail the csv file to [email protected].