Reproduce “Beyond Pixels: Semi-Supervised Semantic Segmentation with a Multi-scale Patch-based Multi-Label Classifier” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Prantik Howlader, Srijan Das, Hieu Le, Dimitris Samaras
MPMC adds patch-level multi-label classification to semi-supervised semantic segmentation, predicting which classes occur in image regions alongside pixel labels. It uses pooled features at two scales and patch confidence to weight teacher-generated supervision. On Cityscapes validation, AugSeg with ResNet-101 and 186 labeled images improves from 75.2 to 76.9 mIoU, the mean per-class intersection-over-union score; this result is limited to that model, dataset, and partition.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.