Dangerous Flight Classification System

Dangerous Flight Classification System of Drone vs. Bird with Alarm Message

Summary of Project and Team

• Microsoft AI School Team 2 Project
• Schedule :
  • Processing : 2023.1.10 ~ 1.17
  • A result briefing session : 2023.1.18 | Presentation : http://naver.me/FBDn8QAO
• Member of Project
  Team Leader : @yeoiksu | https://github.com/yeoiksu
  Team Member : @TaeYounKwon | https://github.com/TaeYounKwon
  Team Member : @ssyjgs1 | https://github.com/ssyjgs1
  Team Member : @Byunggu-Son | https://github.com/Byunggu-Son
  Team Member : @yuyeon | https://github.com/yuyeon-choi
• Project Environment and Technical Stacks
  • Stacks : Python, PyTorch, Anaconda Environment, etc...
  • Environments : Processing Model on Microsoft Azure Data Science Virtual Machine
  • Detail Information of Used Pkg List : package list info.yaml

Introduction

Real-time camera checks drones and birds for binary classification, and when dangerous goods are found, GPS information of dangerous goods is provided to users by email with an alarm message.

Intention of Project

Architecture of System

• We used MobileNetV2 on our system. Because of
  • Accuracy not inferior to other image classification models
  • Requirement of lightweight models in limited environments

Examples of Demonstration Picture