Problem
Manual container inspection is slow and inconsistent, while port environments create visual complexity for detecting dents, holes, rust, and other subtle structural damage.
Back to research
ICCIT 2025
AutoContDamage: Automated Container Damage Detection in Ports Using RF-DETR
A conference paper by A. Rahman, M. S. H. Khan, M. Tazwar, M. T. Rahat, and K. Nur on RF-DETR-based damage detection for port logistics.
Why It Matters
Better automated inspection can reduce delays, minimize human error, and improve operational efficiency in containerized logistics.
Core Idea
The paper applies RF-DETR to container damage detection and uses augmentation to improve detection of small or visually subtle defects in realistic port conditions.
Main Result
Indexed metadata reports RF-DETR performance with 94.1% mAP, 92.6% precision, and 90.0% recall.
What Was Difficult
The hard cases are small defects, cluttered backgrounds, occlusion, corrosion patterns, and visual variation across real-world shipping containers.
Resources
Open Directions
- Can RF-DETR generalize to rare damage types with limited examples?
- How should container damage datasets represent port lighting and occlusion?
- Can inspection models be deployed in a low-latency port workflow?