AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

FocusDiffuser: Perceiving Local Disparities for Camouflaged Object Detection

Jianwei Zhao, Xin Li, Fan Yang, Qiang Zhai, Ao Luo, Zhicheng Jiao, Hong Cheng

ECCV 2024 · Detection and segmentation › Open-set and few-shot detection · approach: Open-set and unknown object detection · 29 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
replace transformer decoder with diffusion model (ablation)FocusDiffuser (PVTv2) · COD10Kweighted F-measure: 0.673 -> 0.787Improved
add BDLU module (ablation)diffusion baseline · COD10KMAE: 0.033 -> 0.027Improved
add Cyclic Positioning module (ablation)diffusion baseline · COD10KMAE: 0.033 -> 0.022Improved
combine BDLU and CP (ablation)diffusion baseline · COD10KMAE: 0.033 -> 0.020Improved
add correlation LookUp without boundary selection (ablation)BDLU · COD10KS-alpha: 0.869 -> 0.867Hurt
add boundary selection to lookup (ablation)BDLU · COD10KS-alpha: 0.867 -> 0.875Improved
increase lookup grid radius from 2 to 4 (ablation)BDLU · COD10KS-alpha: 0.875 -> 0.865Hurt
use bounding-box mask for background filtering in CP instead of all-ones mask (ablation)CP · COD10KS-alpha: 0.867 -> 0.875Improved
increase denoising steps beyond 4 (ablation)FocusDiffuser · COD10KS-alpha: 0.875 -> 0.875 (N=8)No clear effect
reduce denoising steps from 4 to 2 (ablation)FocusDiffuser · COD10KS-alpha: 0.875 -> 0.870Hurt
lower signal scale factor b from 1.0 to 0.1 (ablation)FocusDiffuser · COD10KS-alpha: 0.859 -> 0.875Improved
lower scale factor b further from 0.1 to 0.05 (ablation)FocusDiffuser · COD10KS-alpha: 0.875 -> 0.870Hurt

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