AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Explanations for Occluded Images

Hana Chockler, Daniel Kroening, Youcheng Sun

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 25 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
use causal-responsibility explanations instead of SBFL-based rankingDC-Causal · MS-COCO partially occluded imagesexplanation size (% of pixels): 37.8 -> 13Improved
use causal ranking instead of DC-SBFL when keeping 20% of pixelsDC-Causal · MS-COCO partially occluded imagescorrect classification (%): 80% correct; more than 20% improvement over DC-SBFLImproved
use DC-Causal instead of RISEDC-Causal · Photo Bombingexplanations with no occlusion overlap: more than 60%, 20% better than RISEImproved
vary the masking color from black to white (ablation)DC-Causal · Photo Bombingmean intersection with occlusions: influence less than 4%No clear effect
use DC-Causal instead of RISE on explanation sizeDC-Causal · Photo Bombingexplanation overlap with occlusion: smaller explanations but higher overlap with occlusions than RISEMixed

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