ai-monitoring

Features

• Capture screen activity at custom intervals.
• Analyze screen captures for distracting activities using AI models.
• Display visual and audio alerts for detected distractions.
• Track statistics of distractions and activities.
• Customizable activities and alert settings.

Installation & Setup

Step 1: Clone the Repository

Clone the repository to your local machine:

git clone https://github.com/DerekLiu35/ai-monitoring.git
cd ai-monitoring

Step 2: Dependencies Installation

1. Python 3.7+
2. PIP (Python package manager)

To keep your project dependencies organized and avoid potential conflicts with other Python projects, it's recommended to use a Python virtual environment (venv) (or conda). Here’s how to set up and activate a virtual environment before installing dependencies.

In the project directory, run:

python3 -m venv venv

This will create a new folder named venv containing a local Python environment.

Then,

• On macOS/Linux:

  source venv/bin/activate

• On Windows:

  .\venv\Scripts\activate

After activation, you’ll see (venv) at the beginning of your command prompt, indicating that the virtual environment is active.

With the virtual environment activated, install the project dependencies using:

pip3 install -r requirements.txt

Step 3: Model Selection and Configuration

This application uses AI models to analyze screen captures and determine if the user is engaged in distracting activities. The current setup supports two model options:

1. Gemini Model:

   • Model Name: gemini-1.5-flash-latest
   • This model is available via the Google Gemini API. For more accurate analysis, Gemini is recommended.
   • Configuration: Add your Gemini API key to the .env file as follows:

     GEMINI_API_KEY=your_gemini_api_key_here
     

2. LLaVA Model via Ollama:

   • Ollama provides a straightforward way to run LLaVA on a local machine without needing an external API, enhancing privacy.
   • Setup: Install Ollama at Ollama's official site. Once installed, run Ollama to serve LLaVA as follows:

     ollama run llava

   • Endpoint: Ollama will create an endpoint at http://localhost:11434, which the application uses to interact with the LLaVA model.

Step 4: Run the Application

To start the application, execute:

python app.py

Step 5: Usage

1. Set the interval, activities, and blacklisted words in the settings.
2. Click "Start Monitoring" to begin.
3. The app will periodically capture the screen, analyze activities, and alert you if you’re distracted.

Additional Notes

• Audio Alerts: Ensure the Radar.mp3 file (or your custom audio) is available in the directory specified in config.json.

Troubleshooting

• Screen Capture Issues: Ensure you have the correct permissions and screen capture software.
• Audio Alerts: Ensure audio files are accessible and compatible with playsound.