threat-scanner 0.1.1

ThreatScan is an open-source Python package designed to detect potential physical threats in videos

ThreatScan

ThreatScan is an open-source Python package designed to detect potential physical threats in videos. It leverages the power of video masking autoencoder (VideoMAE) classification models, fine-tuning them to specifically identify events such as fires and other user-defined threats.

Overview

This package provides a streamlined approach to:

• Utilize pre-trained VideoMAE models: Benefit from state-of-the-art video understanding capabilities.
• Fine-tune for threat detection: Adapt the base model to accurately classify specific threat events present in video data.
• Scalable threat analysis: Process video streams or individual video files for real-time or batch analysis.
• Extensible architecture: Easily integrate custom threat classes and extend the model's capabilities.

Installation

You can install ThreatScan using pip:

pip install threatscan

Getting Started

This document outlines the setup and usage instructions for the project.

Setting up your Virtual Environment

It's highly recommended to use a virtual environment to isolate project dependencies. Follow these steps:

1. Create a virtual environment:

   python3 -m venv .venv
   

   This command creates a new virtual environment in a directory named .venv within your project.

2. Activate the virtual environment:

   • On macOS and Linux:

     source .venv/bin/activate
     

     Your terminal prompt should now be prefixed with (.venv), indicating that the virtual environment is active.

   • On Windows (Command Prompt):

     .venv\Scripts\activate
     

   • On Windows (PowerShell):

     .venv\Scripts\Activate.ps1
     

Installing Dependencies

Once the virtual environment is activated, install the project's required libraries using pip:

pip install -r requirements.txt

Running the Application

You can run the main application in different modes:

• Live Feed (Default) To run the application using your default webcam or live video feed:

python3 -m main

• With a Video File To run the application using a specific video file:

python3 -m main --source threatscan/detector/examples/fire.mp4

Replace threatscan/detector/examples/fire.mp4 with the actual path to your video file.

• With a Custom Model To run the application using a specific trained model:

python3 -m main --model mymodel/path/etc

Replace mymodel/path/etc with the correct path to your model file.

Training your own model

To train your own data go to Training README

Contributing

Contributions are welcome! Please fork the repository and submit a pull request with your improvements.

License

This project is licensed under the MIT License. See the LICENSE file for details.