A Vision Transformer Model-Based Shipping Container Damage Inspection Scheme

This paper proposes a scheme for inspecting containers for aging and damage using the Vision Transformer (ViT) model. The long-term use of port machinery and repeated handling can cause containers to age and become damaged. To address this problem, we apply the ViT model to accurately identify various aging conditions and types of damage, replacing the need for manual inspections by freight personnel. Our method uses computer vision techniques to capture local and global detailed information in container aging and damage images, thereby improving the accuracy of container condition detection. The proposed method achieved an accuracy of 80.6%, a loss function of 0.724, and a learning rate of 0.001 based on experimental results.