Smart Captcha API

This FastAPI-based application detects if interactions with a website are coming from a human or a bot, based on mouse and keyboard behavior. It also uses browser information and device fingerprints to assess if the client is automated or previously blacklisted.

Features

• Mouse and Keyboard Behavior Analysis: The API uses pre-trained models to predict if the interaction is from a human or a bot based on mouse and keyboard movement patterns.
• Browser Automation Detection: By analyzing browser data, the system checks for anomalies in userAgent, platform, and other properties to determine if the browser is automated.
• Device Fingerprinting: The system logs device fingerprints to Firebase Realtime Database and flags devices as bots if they make frequent requests.
• Firebase Realtime Database: Stores visit information, device fingerprints, and bot statuses using Firebase's real-time database.

Requirements

• Python 3.7+
• FastAPI
• Uvicorn
• Firebase Admin SDK
• Joblib
• Pandas
• NumPy
• Scikit-learn

Python Packages

Install the following packages using pip:

pip install -r requirements.txt

Setup

1. Move to the App Directory:

   • Move to the directory using cd:

   cd app

2. Start the Server:

   • Run the FastAPI server using Uvicorn:

   uvicorn main:app --reload

   This will start the API server locally at http://127.0.0.1:8000.

API Endpoints

/predict_behavior (POST)

Predicts if the user behavior comes from a human or a bot based on mouse, keyboard interactions, and browser information.

Parameters:

• mouse_file: CSV file containing mouse movement data (x, y, timestamp).
• key_file: CSV file containing keyboard event data (fieldName, timestamp).
• browser_info: JSON string of browser details (e.g., userAgent, platform, etc.).
• fingerprint: String representing the device fingerprint.

Response:

{
  "mouse_result": "Human" or "Bot",
  "key_result": "Human" or "Bot",
  "is_automated": "Yes" or "No",
  "is_bot": "Yes" or "No"
}

/add_visit_info (POST)

Logs visit information based on a device fingerprint and timestamp.

Parameters:

• fingerprint: String representing the device fingerprint.
• timestamp: Integer representing the time in milliseconds.

Response:

{
  "message": "Fingerprint and visit info saved successfully!",
  "timestamp": <timestamp>,
  "fingerprint": <fingerprint>
}

Models

The API uses two machine learning models:

1. MouseVerifier.pkl: Detects if mouse movements are human or bot-controlled.
2. KeyboardVerifier.pkl: Detects if keyboard inputs are from a human or a bot.

Ensure these models are stored in the project directory.

Example Usage

To test the /predict_behavior endpoint, use a tool like curl or Postman to send a POST request with the required files and form data.

Example using curl:

curl -X POST "http://127.0.0.1:8000/predict_behavior" \
-F "mouse_file=@path/to/mouse_data.csv" \
-F "key_file=@path/to/key_data.csv" \
-F "browser_info={\"userAgent\": \"Mozilla/5.0 ...\", \"platform\": \"Win32\", ...}" \
-F "fingerprint=abc123"

---