Model weights and data for paper "Self-Supervised Learning Approach for Multi-label Sewer Defect Classification"
DOI:10.4121/1c21ce33-715f-4ca0-89fa-c170b30801ff.v2
The DOI displayed above is for this specific version of this dataset, which is currently the latest. Newer versions may be published in the future.
For a link that will always point to the latest version, please use
DOI: 10.4121/1c21ce33-715f-4ca0-89fa-c170b30801ff
DOI: 10.4121/1c21ce33-715f-4ca0-89fa-c170b30801ff
Datacite citation style
Yıldızlı, Tuğba; Jia, Tianlong; Langeveld, Jeroen; Taormina, Riccardo (2025): Model weights and data for paper "Self-Supervised Learning Approach for Multi-label Sewer Defect Classification". Version 2. 4TU.ResearchData. dataset. https://doi.org/10.4121/1c21ce33-715f-4ca0-89fa-c170b30801ff.v2
Other citation styles (APA, Harvard, MLA, Vancouver, Chicago, IEEE) available at Datacite
Dataset
Version 2 - 2025-11-11 (latest)
Version 1 - 2025-11-04
Repository for model weights and data accompanying the paper “Self-Supervised Learning Approach for Multi-label Sewer Defect Classification” by Tugba Yildizli, Tianlong Jia, Jeroen Langeveld, and Riccardo Taormina.
This repository provides:
- SwAV pre-trained weights for self-supervision,
- Fine-tuned model weights (fully supervised and semi-supervised),
- Supporting data/configs used to train and analyze these models.
Researchers can (i) fine-tune the SwAV pre-trained backbones on their own sewer datasets for semi-supervised learning, and (ii) evaluate our fine-tuned models for reproducibility. All code examples use PyTorch.
History
- 2025-11-04 first online
- 2025-11-11 published, posted
Publisher
4TU.ResearchDataFormat
.torch files for SwAV pre-training weights, .pth files fine-tuned model weigths, .txt files for image names, .csv files for data annotationDerived from
Data link
https://vap.aau.dk/sewer-ml/Organizations
TU Delft, Faculty of Civil Engineering and Geosciences, Department of Water ManagementDATA
Files (2)
- 1,601,860 bytesMD5:
27186bc9cca7f1691a6a69757c5d84caDataset.zip - 5,299,662,200 bytesMD5:
90342804eebf47784aaf118922bd1295Model weights.zip -
download all files (zip)
5,301,264,060 bytes unzipped





