A 2017-era graveyard of semantic segmentation papers

What it does This repository collects links to semantic segmentation papers (mostly 2017–2019), dataset URLs, and a handful of survey resources. It also claims “some implementation of semantic segmantation for DL model,” though the README is overwhelmingly a bibliography with paper links and occasional code references.

The interesting bit The dataset list is almost comically comprehensive—spanning from PASCAL VOC to liver tumor segmentation to “Data from Games”—making it a decent starting point for hunting down niche training data. The single embedded image, a “Dataset importance plot” from an external domain, hints at ambitions of analysis that the rest of the repo doesn’t deliver.

Key highlights

• Heavy focus on 2017–2019 CVPR/ICCV/ECCV papers (DeepLab v3/v4, ICNet, SegNet era)
• Extensive dataset index: 25+ sources including medical, autonomous driving, and scene parsing
• Links to survey papers and two live demos (CRF-as-RNN, SegNet)
• Sparse “Resources” and “Survey papers” sections with minimal curation
• Jupyter Notebook language tag suggests some implementations exist, but README doesn’t surface them clearly

Caveats

• README is truncated mid-paper-list in both provided sources, so completeness is unclear
• Many paper entries are just titles with links; no summaries, no reproduction notes
• “Some implementation” claim is vague—actual code location and quality are unspecified
• Effectively frozen in pre-transformer, pre-SAM archaeology; nothing from 2020 onward

Verdict Worth a bookmark if you’re doing historical research or need a quick dataset directory for legacy methods. Skip it if you want working code, modern architectures, or curated guidance—this is a link dump with delusions of grandeur.