Important

THIS IS THE V3-MODEL ENHANCED VERSION THAT INCLUDES FASTER GRAPHING ALONG WITH OTHER IMPROVEMENTS! This version has the best version of the eye tracking technology integrated with the fastest graphing module. All exports (graph png, csv and mp4) work well.

Note

To create an environment capable to running this on your local device, there is now conda yaml support for dependency management.

DETECT: Deception Tracking Through Eye Cues Technology

Note: This project has just been started and is mostly research as of now.

Welcome to DETECT! This project aims to revolutionize the field of deception detection by leveraging advanced gaze tracking technology. Utilizing the powerful MediaPipe framework, DETECT analyzes eye cues to offer a new dimension in understanding and interpreting human behavior.

📜 Overview

DETECT (Deception Tracking Through Eye Cues Technology) is a cutting-edge project designed to track and analyze eye movements to assist in identifying potential deception. By triangulating the position of the iris, DETECT provides valuable insights into gaze patterns that can be indicative of truthfulness or deception.

⚙️ Features

The following features are currently available (almost all are experimental :P):

Feature	Description
affine	Use the Face normalization algorithm for possible improvement in precision
graph	Graph the x-time and y-time plots to see the changes in real time
dot_display	Show the iris/pupil as tracked by mediapipe (might reduce load)
export::csv	Export the tracked eye data into a csv file for advanced analysis
csv_interval [sec]	🚨 [Doesn't work] Sets the time interval between each collected data point for the csv
export::graph	Export the tracked eye data and graph it for easier comprehension and basic analysis
export::animation	Estimate the gaze direction over time based on the tracked eye data
dot_display	Toggle to either have drawing on video output or not
categorize	Requires dot_display to be true; Displays an estimate on which direction of the tracked gaze

---

More features will be coming soon...

🛠️ Installation

To get started with DETECT, follow these steps:

1. Clone the Repository:

   git clone https://github.com/bingKegeta/DETECT.git
   cd DETECT

2. Set Up a Virtual Environment:

   • Using conda/miniconda (recommended):

   conda env create -f environment.yaml

   This will also take care of the dependencies and you can directly start using the program!

   • Using venv:

   python -m venv env
   source env/bin/activate  # On Windows, use `env\Scripts\activate`

   • Docker: 👨‍🍳🍳
   • Others: Task will be left to the reader

3. Install Dependencies (venv and other pip-based package managers):

   pip install -r requirements.txt

🔧 Usage

1. Make a configuration file (JSON):

   • The file contains various options but the basic schema is this:

   {
      "source":["webcam", "image", "video"], # Choose any one of those
      "path": "/path/to/media", # Only include when the source is not webcam
      "graph": boolean, # Only if you want graph updates (fps will be hit)
      "affine": boolean, # If you want affine tranform processing
      "csv_interval": number, # Doesn't work yet
      "export": {
         "csv": boolean, # Do you want csv output?
         "graph": boolean, # Do you want a comprehensive graph?
         "animation": boolean, # Do you want a nifty animation?
      },
      "export_dir": "/path/to/dir/" # Where do you want to store those files?
   }

2. Run the Application:

   python main.py config.json

3. Start Analyzing:

   • The application will initiate your camera and begin tracking eye movements.
   • Use the provided interface to view and analyze gaze data.

📜 License

This project is licensed under the MIT License.

---