Automated sewer defect detection has advanced through deep learning, particularly supervised methods using CCTV images, but based on large annotated datasets. This paper proposes a semi-supervised learning (SSL) approach to reduce labeling demands. The method comprises self-supervised pre-training on unlabeled images using SwAV (Swapping Assignments between multiple Views) followed by fine-tuning for multi-label classification. Experiments on the Sewer-ML dataset demonstrate that the SSL approach, trained on only 35k labeled images, achieves an F1-score of 69.11%, and F2_CIW of 54.22%, surpassing the fully supervised baseline trained from scratch on 1.04 million images. Increasing the unlabeled pre-training data further enhances performance, while ImageNet initialization consistently outperforms training from scratch. Self-supervised learning also helps mitigate the effects of mislabeled data, which is observed to be present even in the Sewer-ML ground truth. Overall, self-supervised learning provides an accurate, scalable, and cost-effective alternative to fully supervised approaches, particularly in data-scarce or imperfectly labeled scenarios.

This opinion paper reflects on the current challenges facing urban drainage systems (UDS) research, along with solutions for fostering sustainable development. Over the course of a year-long project involving 92 participants aged 24-38, including PhD candidates, post-doctoral researchers, and early-career academics, we identified critical challenges and opportunities for the sustainable development of UDS. Our exploration highlights four key challenges: limited public visibility leading to resource constraints, insufficient collaboration across subfields, issues with data scarcity and data sharing, and geographical specificities. We emphasise the importance of raising public and political awareness regarding UDS's vital role in climate adaptation and urban resilience, advocating for blue-green infrastructure and open data practices. Additionally, we address systemic academic barriers that hinder innovative research. We call for a shift away from metrics that prioritise quantity over quality. We recommend establishing stable career pathways that empower early-career researchers. This paper aims to catalyse a broader community dialogue about the future of UDS research, uniting voices from various career stages. By presenting actionable recommendations, we aim to inspire fundamental changes in research conduct, evaluation, and sustainability, ensuring the field of UDS is prepared to meet pressing urban water management challenges worldwide.