Skip to content

boostcampaitech2/final-project-level3-cv-06

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

X-Ray Baggage Scanner 자동 검출 솔루션

강지우 곽지윤 서지유 송나은(NaeunSong) 오재환 이준혁 전경재

Project overview

현재 사람이 직접 위해물품을 탐지하는 시스템에서는 human inspection error로 인해 80~90%의 정확도를 보입니다. 그리하여 AI 기술을 활용한 X-Ray Baggage Scanner 자동 검출 솔루션을 통해 더 높은 정확도를 보장하고 다양한 품목에 대하여 검출 가능하도록 하였습니다.

Project roadmap

그림2

Dataset

Contents

Root
├── app
├── data
├── detection
├── Dockerfile
├── Makefile
├── multilabel
├── README.md
├── requirements.txt
├── setup.py
└── utils
    └── timer.py
  • app/ : web resource
  • data/ : train data path (mount dataset path)
  • detection/ : detection model path
  • multilabel/ : multilabel classification model path
  • Makefile : Web Server Entrypoint
  • Dockerfile : Dockerfile

Result

  • 기존 20개 위해품목 감지
    • 38개 전체 위해품목 감지 가능
  • 기존 정확도 90% 이상
    • (Multilabel)EMR 기준 0.99 이상
    • (Object detection) mAP 기준 0.9 이상
  • 기존 inference time 0.7s/image
    • under 0.2s/image (with GPU)
    • under 5s/image (without GPU)
      • GPU 추가 없이 소프트웨어 설치만으로 이용 가능한 수준

Test setting

  • CPU : Intel(R) Xeon(TM) i7-8700 CPU(12 core)
  • GPU : GeForce GTX 1080

Train setting

  • CPU : Intel(R) Xeon(R) Gold 5120 CPU @ 2.20GHz(8 core)
  • GPU : Tesla V100 32gb

About

final-project-level3-cv-06 created by GitHub Classroom

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published