Drone Surveillance System

Keeping in view the security aspects when related to crowds in public, it is often difficult to monitor and safeguard areas of potential threats. With the recent advancement in technology, Drone systems have been widely deployed by various law enforcement agencies to monitor areas for public sa

Project Title

Drone Surveillance System

Project Area of Specialization

Artificial Intelligence

Project Summary

Keeping in view the security aspects when related to crowds in public, it is often difficult to monitor and safeguard areas of potential threats. With the recent advancement in technology, Drone systems have been widely deployed by various law enforcement agencies to monitor areas for public safety. In this project, we propose a real-time drone surveillance system to identify violent individuals in public areas.  The project provides autonomous security protection in public crowds to accurately detect anomalies in the crowd and alert the security action individual in real-time before anything goes wrong and the situation can be handled properly. A twofold system is designed based on Android and Windows systems. The Android app simply classifies the video frames as Violent or Non-violent classes while the windows app localizes the violent individual for more detailed inspection.

Project Objectives

The project is majorly aimed to achieve the following objectives:

- Provide an efficient and low-cost solution to crowd monitoring.

-  Identify violent individuals in the crowd in real-time.

-  Provide a reliable and a portable system

- Exploit the power of AI to monitor crowds

Project Implementation Method

The system first uses the YOLOv4 architecture to detect humans from aerial video frames. The humans are then tracked in multiple frames resulting in the same ids for the same people in multiple frames. A queue is maintained containing position coordinates for every individual person id in 20 frames. The queue is then used by the proposed Neural Network consisting of convolutional layers for feature extraction and LSTM to identify the violent individuals. The dataset consists of almost 2000 videos having two different sets i.e. violent and non-violent activity.

Benefits of the Project

By the integration of our proposed work, the user can be benefited in the following ways:

1. Cost Saving: instead on installing dozens of CCTV cameras at a variety of spots, the user can simply position a single drone and move it along with the crowd. 

2. Real-time identification

3. Crowd monitoring solution

Technical Details of Final Deliverable

The final deliverables include:

- An android application copped in with the capability to seamlessly communicate with the drone and receive a video feed that is processed and classifies the frame as either violent or non-violent on the basis of our proposed deep learning neural network.

- A windows application is able to spot the violent individual in the crowd by the means of bounding boxes and person id.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Security

Other Industries

Education , Media

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT)

Sustainable Development Goals

Decent Work and Economic Growth, Industry, Innovation and Infrastructure, Peace and Justice Strong Institutions, Partnerships to achieve the Goal

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com