Deep Learning Based Human Detection and Tracking Drone

Over the past few years, the use of drones has rapidly grown in popularity offering multipurpose uses, the drones have become an important focus in various applications including agricultural monitoring, disaster management, surveillance, remote sensing, target acquisition, border patrol, infrastruc

Project Title

Deep Learning Based Human Detection and Tracking Drone

Project Area of Specialization

Artificial Intelligence

Project Summary

Over the past few years, the use of drones has rapidly grown in popularity offering multipurpose uses, the drones have become an important focus in various applications including agricultural monitoring, disaster management, surveillance, remote sensing, target acquisition, border patrol, infrastructure monitoring, photography, and videography. Moreover, drones are a useful tool for researchers to test and evaluate new ideas in various fields, including flight controls theory, navigation, real-time systems, and robotics.
In this project, a deep learning-based Real-time Human detection and tracking are proposed. In the first phase, a custom deep learning framework will be designed and trained using the Stanford Drone dataset. 
In the second phase, advanced computer vision technology shall be employed to track the particular human.
Initially, the proposed techniques will be validated using simulations and then real-time flights on a quadcopter will be carried out.

And this project will be really very helpful for all our security institutions like the police and intelligence agencies in identifying and tracking a suspicious person. And it will also help in rescue operations in difficult areas like forests etc.

Project Objectives

1) Develop a deep learning-based human detection algorithm using Pytorch.

2) Develop a deep learning-based human tracking algorithm using Pytorch.

3) Setup the hardware, mount the gimbal and camera to the drone

4) Implementation of the developed algorithms on Drone for real-time test flights.

Project Implementation Method

Hardware

• Purchasing of hardware components based on design requirements
• Setting Up the Hardware components and mounting to the drone
• Integration of camera for live video streaming
• Transmission of a video stream to display via video transceiver/router

Software

• Deep Learning-based human detection using YOLO/Retinanet framework
• Feature-based tracking using real-time tracker DEEPSORT
• Training of the network using Stanford Drone Dataset
• Testing the simulations and improving the network parameters for desired results

Hardware and software integration

• Implement the developed network architecture on the Drone's Jetson Board

Benefits of the Project

It can be used in the detection of humans in a given area and tracking a particular selected human. Law enforcing agencies can use for aerial surveillance of particular areas at the borders in case of an enemy or an intruder. Vloggers can use this for aerial video shots including the tracking feature.

Technical Details of Final Deliverable

A machine learning/ deep learning-based human detection & tracking algorithms will be designed and implemented on a Jetson Tx1 and real-time test flights of quadcopter will be conducted.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Security

Other Industries

Others

Core Technology

Artificial Intelligence(AI)

Other Technologies

Robotics

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

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