Soft Proposal Networks for Weakly Supervised Object Localization

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

  • SPN is an end-to-end weakly supervised localization framework.
  • SPN introduce the nearly cost-free object proposal.
  • SPN learns better object-centric filters and discover more visual evidence
  • SPN is compatible with popular architectures (VGG/GoogLeNet etc.)

SPN is an end-to-end weakly supervised localization framework.

Source code and Demo Paper (arXiv) Check Yi's Awesome Page for More Resources