Intelligent Detection of Container Surface Damage Based on Scene Self-Adaptation and Cascaded Network

Abstract

As the core carrier of the global logistics chain, the structural integrity of the outer surface of the container directly determines the safety of cargo transportation and the efficiency of port operations. During long-term use, containers are prone to seven common types of damage, such as holes, damage, and rust, due to factors such as mechanical collisions, environmental corrosion, and structural fatigue. Traditional artificial vision detection methods are inefficient and subjective, making it difficult to meet the operational needs of modern port automation and high throughput. The rise of Computer Vision and Deep Learning technology provides an effective solution for intelligent detection of container damage.This paper focuses on the automatic detection and classification of seven types of damage on the outer surface of containers. In response to core challenges such as complex natural lighting, visually similar damage, and non-standard data annotation, it constructs an efficient and intelligent detection system to facilitate the automated upgrade of port inspections.