AI Paper Atlas

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

Complementary Patch for Weakly Supervised Semantic Segmentation

Fei Zhang, Chaochen Gu, Chenyue Zhang, Yuchao Dai

ICCV 2021 · Detection and segmentation › Weakly supervised detection and segmentation · approach: CAM refinement and prototypes · 161 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 L_tcp and PCM (ablation)baseline CAM · VOC12 train pseudo-label mIoUmIoU (%): 47.84 -> 51.08Improved
add CP cross regularization L_cpcr (ablation)CPN · VOC12 train pseudo-label mIoUmIoU (%): 51.08 -> 55.71Improved
apply online hard example mining to L_cpcr (ablation)CPN · VOC12 train pseudo-label mIoUmIoU (%): 55.71 -> 56.58Improved
add Pixel-Region Correlation Module (PRCM) (ablation)CPN · VOC12 train pseudo-label mIoUmIoU (%): 56.58 -> 57.43Improved
add PRCM to SEAM (multi-scale) (ablation)SEAM · VOC12 train pseudo-label mIoUmIoU (%): 55.41 -> 55.71Improved
use CPN instead of baseline CAM (multi-scale)baseline CAM · VOC12 train pseudo-label mIoUmIoU (%): 47.84 -> 57.43Improved
use CPN instead of SEAM at image scale 0.5SEAM · VOC12 train pseudo-label mIoU (scale 0.5)mIoU (%): 49.35 -> 48.91Hurt
use CPN instead of SEAM (multi-scale)SEAM · VOC12 train pseudo-label mIoUmIoU (%): 55.41 -> 57.43Improved
increase super-pixel number from 200 to 8000 (smaller patches) (ablation)CPN super-pixel patch · VOC12 train pseudo-label mIoUmIoU (%): 57.43 -> 55.79Hurt
use grid patch set K={56,112} instead of K={4,7} (ablation)CPN grid patch · VOC12 train pseudo-label mIoUmIoU (%): 55.67 -> 57.07Improved
use grid patch set K={224,448} instead of K={56,112} (larger patches) (ablation)CPN grid patch · VOC12 train pseudo-label mIoUmIoU (%): 57.07 -> 55.80Hurt
raise hidden probability from 0.1 to 0.5 (ablation)CPN · VOC12 train pseudo-label mIoUmIoU (%): 55.52 -> 57.43Improved

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