This repository contains the source code and supplementary materials for my undergraduate thesis in Automation Engineering at FURG. The project explores the use of semi-supervised learning techniques for segmenting subsea pipelines in underwater images, focusing on enhancing performance in scenarios with limited labeled data.
Key features:
Implementation of state-of-the-art semi-supervised segmentation models.
Dataset preparation and preprocessing for underwater images.
Comparative analysis of model performance with traditional and modern techniques.
Detailed documentation and scripts for training and evaluation.
Feel free to explore, contribute, or adapt the code for similar projects in underwater imaging and automation!