OCR 81mm 기반 박격포탄 표기 상태판별OCR-Based 81 mm Mortar Bullet Marking Status Judgment
- Other Titles
- OCR-Based 81 mm Mortar Bullet Marking Status Judgment
- Authors
- 윤정환; 이선우; 김규철; 신백천; 허장욱
- Issue Date
- Jul-2024
- Publisher
- 한국기계가공학회
- Keywords
- 81mm Mortar Shell(81mm); OCR Technology(OCR 기술); Data Preprocessing(전처리); Performance Evaluation(성능평가); Automation(자동화)
- Citation
- 한국기계가공학회지, v.23, no.7, pp 17 - 23
- Pages
- 7
- Journal Title
- 한국기계가공학회지
- Volume
- 23
- Number
- 7
- Start Page
- 17
- End Page
- 23
- URI
- https://scholarworks.bwise.kr/kumoh/handle/2020.sw.kumoh/28866
- DOI
- 10.14775/ksmpe.2024.23.07.017
- ISSN
- 1598-6721
2288-0771
- Abstract
- This study introduces an automated system using optical character recognition (OCR) technology to identifymarkings on 81-mm mortars, aiming to overcome the inefficiencies and inconsistencies of expensive andtime-consuming manual inspection processes. We created a measurement environment to identify the markingsof 81-mm mortar shells and trained and evaluated their performance using the representative model ofperformance and EasyOCR framework using multilingual text detection (MLT). The model was trained usingthe AI-HUB dataset and self-generated data as training data, with its performance evaluated using precision,recall, and F-measures. The experimental results indicated that the EasyOCR model performed similarly torepresentative models. Therefore, the automated system proposed in this study can effectively identify themarkings of 81-mm mortars and is expected to contribute toward improving the efficiency and reliability ofthe inspection and discrimination process, thus enabling efficient management and maintenance of mortars—essential for military operations.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - School of Mechanical System Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/kumoh/handle/2020.sw.kumoh/28866)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.