Synthetic Image Datasets of Port Fenders Generated Using LoRA-Based Diffusion Models

This dataset contains synthetic images of port fenders generated using Low-Rank Adaptation (LoRA) models applied to Stable Diffusion. The datasets were developed as part of a research project at the Federal Institute of Education, Science and Technology of Maranhão (IFMA), Brazil, focused on data augmentation techniques for defect detection in ship fenders installed at the Port of Itaqui. Two distinct collections of synthetic images are provided. They correspond to different image generation allowing researchers to explore and compare the visual outcomes obtained from different training data configurations. The broader research investigated the use of synthetic image generation to address the limited availability of real-world images of ship fender defects, including structural damage and corrosion. Research institution: Federal Institute of Education, Science and Technology of Maranhão (IFMA), Brazil.