AI Paper Atlas

Research map / Detection and segmentation / CNN features

RGB-D Saliency Detection via Cascaded Mutual Information Minimization

Jing Zhang, Deng-Ping Fan, Yuchao Dai, Xin Yu, Yiran Zhong, Nick Barnes, Ling Shao

ICCV 2021 · Detection and segmentation › CNN features · approach: RGB-D salient object detection · 146 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 mutual information minimization regularizer (lambda 0 -> 0.1) (ablation)CMINet · NJU2KS-measure: .918 -> .939Improved
add cascaded MI regularizer framework over the plain baseline (ablation)CMINet · NJU2KS-measure: .910 -> .939Improved
add MI regularizer: feature embedding cosine similarity (NLPR) (ablation)CMINet · NLPRcosine similarity of z_a, z_g: 0.90 -> 0.11Improved
use embedding dimension K=6 instead of K=3 (ablation)CMINet · NJU2KS-measure: .928 -> .939Improved
use embedding dimension K=6 instead of K=32 (ablation)CMINet · NJU2KS-measure: .924 -> .939Improved
use conv plus fc embedding instead of two fc layers (ablation)CMINet · NJU2KS-measure: .926 -> .939Improved
use regularizer weight 0.1 instead of 1 (ablation)CMINet · NJU2KS-measure: .919 -> .939Improved
add final fusion of P_a, P_g, P_f into P (ablation)CMINet · NJU2KS-measure: .925 -> .939Improved
add depth input (RGB-only to early-fusion RGB-D) (ablation)CMINet encoder-decoder · NJU2KS-measure: .898 -> .915Improved
polygon instead of scribble annotation supervisionCMINet weak baseline · COME15K-NormalS-measure: .802 -> .827Improved

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