Weakly Supervised Instance Segmentation using Class Peak Response

Yanzhao Zhou, Yi Zhu, Qixiang Ye, Qiang Qiu, and Jianbin Jiao

  • PRM extracts fine-detailed instance-aware visual cues from class peak responses
  • PRM enables classification networks for instance mask extraction
  • PRM is naturally compatible with modern architectures (VGG/ResNet etc.)
  • PRM introduces negligible training overhead

PRM extracts fine-detailed instance-aware visual cues from class peak responses

Source code and Demo Paper (arXiv) Supplementary materials Poster Presentation