AI Paper Atlas

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

Supervision by Fusion: Towards Unsupervised Learning of Deep Salient Object Detector

Dingwen Zhang, Junwei Han, Yu Zhang

ICCV 2017 · Detection and segmentation › Weakly supervised detection and segmentation · approach: Weakly supervised detection · 173 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 superpixel-level fusion map supervision (SFM) (ablation)OURS · ECSSDAP: .852 -> .889Improved
add superpixel-level fusion map supervision (SFM) (ablation)OURS · DUT-OAP: .640 -> .715Improved
add image-level fusion map supervision (IFM) (ablation)OURS · ECSSDAP: .844 -> .889Improved
add image-level fusion map supervision (IFM) (ablation)OURS · DUT-OAP: .649 -> .715Improved
add superpixel-level confidence map (b) (ablation)OURS · ECSSDAP: .864 -> .889Improved
add superpixel-level confidence map (b) (ablation)OURS · PASCAL-SAP: .759 -> .791Improved
add image-level confidence weight (beta) (ablation)OURS · ECSSDAP: .876 -> .889Improved
add image-level confidence weight (beta) (ablation)OURS · DUT-OAP: .677 -> .715Improved
add image-level confidence weight (beta) (ablation)OURS · SODAP: .754 -> .789Improved
add all four supervisory signals; average drop when removing each (ablation)OURS · all four datasetsaverage drop over metrics: SFM (0.4839) > IFM (0.431) > b (0.312) > β (0.186)Improved
use fused unsupervised supervision instead of the best single unsupervised detectorDHSNet-style detector · all four datasetsAP: more than 10% in terms of the AP scoreImproved
train with human annotation (DHS) instead of fusion pseudo labelsOURS · ECSSDAP: .889 -> .916Improved

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