Fidi Wincoko Putro
Center of Excellence for Motion Technology for Safety Health and Wellness, Telkom University, Indonesia

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

A VGG16-Based CRAFT Network for Vertical Container Number Detection Pima Hani Safitri; Alqis Rausanfita; Fidi Wincoko Putro
Indonesian Journal of Computing, Engineering, and Design (IJoCED) Vol. 8 No. 2 (2026): IJoCED
Publisher : Faculty of Engineering and Technology, Sampoerna University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35806/ijoced.v8i2.612

Abstract

Container identification is critical in logistics and port management, where container numbers must be detected accurately for tracking and operational efficiency. In practice, container numbers may appear in vertical orientations, which challenges conventional optical character recognition (OCR) systems. This study proposes a vertical container-number detection approach based on Character Region Awareness for Text Detection (CRAFT) with a VGG16 backbone. The method focuses on character-level localization using annotated container-number datasets, including images captured in a harbor environment in Surabaya, Indonesia. The CRAFT-based model localizes individual characters under varying orientations, enabling more reliable detection of vertically arranged container numbers. Experimental results show that the method achieved a maximum Intersection over Union (IoU) of 0.944 and an average IoU of 0.389 across the evaluated test images over 10 reported iterations. These findings demonstrate the potential of the proposed approach for vertical container-number detection in real-world port environments and its applicability to automated container-management systems.