AI Paper Atlas

Research map / Detection and segmentation / Weakly supervised detection and segmentation

Weakly Supervised Segmentation of Small Buildings With Point Labels

Jae-Hun Lee, ChanYoung Kim, Sanghoon Sull

ICCV 2021 · Detection and segmentation › Weakly supervised detection and segmentation · approach: Weakly supervised semantic segmentation · 27 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add size-separated loss with small object mask (whole image, no point sampling) over size-weighted U-netRU+wsize · CrowdAI Mapping ChallengeAll F-1: 0.8053 -> 0.8295Improved
add uncertainty point sampling to Ours-I (ablation)Ours-I · CrowdAI Mapping Challengesmall-object recall: 0.4551 -> 0.6068Improved
add uncertainty point sampling to Ours-I (ablation)Ours-I · CrowdAI Mapping Challengesmall-object precision: 0.3132 -> 0.2702Hurt
add uncertainty sampling to Ours-I (ablation)Ours-I · CrowdAI Mapping ChallengeAll F-1: 0.8295 -> 0.8319Improved
add memory bank and false positive sampling to Ours-P-U (ablation)Ours-P-U · CrowdAI Mapping Challengesmall-object precision: 0.2702 -> 0.3607Improved
add memory bank and false positive sampling to Ours-P-U (ablation)Ours-P-U · CrowdAI Mapping ChallengeAll F-1: 0.8319 -> 0.8494Improved
use proposed method instead of ScribbleNetScribbleNet · CrowdAI Mapping ChallengeAll F-1: 0.4895 -> 0.8494Improved
use proposed Ours-P-USB instead of RU+wsizeRU+wsize · WHU buildingAll F-1: 0.8208 -> 0.8490Improved
use proposed Ours-P-USB instead of RU+wsizeRU+wsize · Massachusetts buildingsAll F-1: 0.5413 -> 0.5669Improved
use proposed Ours-P-USB instead of RU+wsize, large-object precision on densely packed buildingsRU+wsize · Massachusetts buildingslarge-object precision: 0.3932 -> 0.4956Improved
use proposed Ours-P-USB instead of RU+wsizeRU+wsize · CrowdAI Mapping Challengesmall-object precision: 0.2593 -> 0.3607Improved
use a smaller point-label radius r (ablation)Ours-P-USB · CrowdAI Mapping ChallengeperformanceHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy