Skip to content

Latest commit

 

History

History
76 lines (52 loc) · 2.21 KB

README.md

File metadata and controls

76 lines (52 loc) · 2.21 KB
Raw

Peye-Viewer Eye Gaze Visualization App

This app is built with Streamlit and visualizes eye gaze data over a video. It overlays real-time eye gaze positions of selected subjects on the video as it plays, allowing you to see how each subject tracks the content. You can toggle individual subjects to visualize their gaze data.

Features

  • Video Player: Plays a video while displaying the corresponding eye gaze data.
  • Subject Selection: Toggle individual subjects to display their gaze data.
  • Gaze Tracking: Shows real-time eye gaze positions as the video plays.

Requirements

The application relies on the following key packages:

  • Streamlit
  • Pandas
  • Numpy
  • video.js (for video rendering)
  • Streamlit Component (for integrating the video player)

Getting Started

1. Clone the repository

git clone https://github.com/cosanlab/peye_viewer.git
cd peye_viewer

2. Setup Conda Environment the repository

You can use Conda to manage the environment for this app. Below are instructions to set up a Conda environment:

a. Create Conda environment

conda create -n peye_viewer python=3.11

b. Activate the environment

conda activate peye_viewer

c. Install dependencies

pip install -r requirements.txt

3. Run the App

Once all dependencies are installed, run the Streamlit app by executing the following command:

streamlit run peye_viewer/viewer.py

4. Video and Data File

Ensure that your video and gaze data files are correctly set up. The paths to these resources are configured in the code:

  • Video URL: Defined in the viewer.py script
  • Eye Gaze Data File: Defined in the DATA_FILE constant in viewer.py

6. Expected Data Format

The CSV data should include at least the following columns for eye gaze data:

  • SID: Subject ID
  • gaze_x: Gaze position on the x-axis (normalized between 0 and 1)
  • gaze_y: Gaze position on the y-axis (normalized between 0 and 1)
  • relative_timestamp: Time in seconds relative to the start of the video

Contribution

Feel free to submit a pull request or open an issue if you encounter any problems or have suggestions.

License

This project is licensed under the MIT License.