Voilence_detection_using_AI

An AI/ML algorithm improves real-time CCTV monitoring by detecting violence percentages and sending alerts to a registered person when violence increases. Secured by Patent Application No: 202341029626.

---

Real-time CCTV Monitoring with Violence Detection

Patent Application No: 202341029626

Overview

This project involves an AI/ML algorithm to enhance real-time CCTV monitoring by detecting and analyzing violent activities within the monitored area. The system calculates the percentage of violence detected and sends immediate alerts to a registered individual when the violence exceeds a certain threshold.

Features

• Real-Time Monitoring: Continuously monitors CCTV feeds for any signs of violence.
• Violence Detection: Utilizes AI/ML models to detect violent actions and behaviors within the feed.
• Percentage Calculation: Calculates the percentage of violence detected in the video footage.
• Alert System: Automatically alerts a registered person when the detected violence percentage exceeds a predefined threshold.
• Secure Implementation: Secured under Patent Application No: 202341029626.

Prerequisites

• Python 3.7+
• TensorFlow or PyTorch (depending on the chosen ML framework)
• OpenCV
• Pre-trained AI/ML models (provided in the repository)
• Access to CCTV camera feeds

Installation

1. Clone the Repository:

   git clone https://github.com/**********/violence-detection-cctv.git
   cd violence-detection-cctv

2. Install Dependencies:

   pip install -r requirements.txt

3. Configure CCTV Feeds:

   • Update the config.json file with the URLs or paths to your CCTV feeds.
   • Register the person who will receive alerts by updating the alert_config.json file with their contact details.

Usage

1. Run the Violence Detection Algorithm:

   python detect_violence.py

2. Monitoring:

   • The system will start monitoring the CCTV feeds in real-time.
   • If violence is detected and the percentage exceeds the threshold, an alert will be sent to the registered person.

3. Logs and Reports:

   • All detected incidents are logged in the logs/ directory.
   • Summary reports can be generated using the generate_report.py script.

Customization

• Violence Threshold: You can adjust the threshold for violence detection in the config.json file.
• Model Tuning: If you wish to train or fine-tune the model, refer to the model_training/ directory.

Contribution

I want you to know that contributions to improve the algorithm or extend its capabilities are welcome. Please fork the repository and submit a pull request with your changes.

License

This project is licensed under the terms of the patent application (No: 202341029626). Commercial use is prohibited without proper authorization.

Contact

For any inquiries or issues, please contact [shinde vinayak rao patil] at [[email protected] ].