AI Paper Atlas

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

Object-Centric Unsupervised Image Captioning

Zihang Meng, David Yang, Xuefei Cao, Ashish Shah, Ser-Nam Lim

ECCV 2022 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 15 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
mine objects from different images instead of pseudo image-sentence pairs (ablation)unsupervised captioner · COCO + SSCIDEr: 0.021 -> 0.039Improved
mine objects from different images instead of pseudo image-sentence pairs (ablation)unsupervised captioner · COCO + GCCCIDEr: 0.025 -> 0.032Improved
raise object coverage from 50.0% to 100% (ablation)object-mining captioner · COCO + SSCIDEr: 0.036 -> 0.039Improved
raise object coverage from 50.0% to 100% (ablation)object-mining captioner · COCO + GCCCIDEr: 0.025 -> 0.032Improved
raise overlap criterion from >= 0 to >= 2 (coverage 4.2% to 47.5%) (ablation)prior pair-mining method · COCO + SSCIDEr: 0.015 -> 0.026Improved
add attribute detector (ablation)object-mining captioner · COCO + LN-OpenImagesCIDEr: 0.109 -> 0.124Improved
add relationship detector (ablation)object-mining captioner · COCO + LN-OpenImagesCIDEr: 0.109 -> 0.120Improved
add both relationship and attribute detectors (ablation)object-mining captioner · COCO + LN-OpenImagesCIDEr: 0.109 -> 0.138Improved
replace objects with same-category objects from other images (ablation)supervised transformer captioner · COCO + LN-COCOCIDEr: 0.263 -> 0.227Hurt
use original locations of substitute objects instead of replaced locations (ablation)supervised transformer with replaced objects · COCO + LN-COCOCIDEr: 0.227 -> 0.239No clear effect
train directly in target language instead of translating English captions (ablation)object-mining captioner · COCO-CN ChineseCIDEr: 0.059 -> 0.349Improved
train directly in target language instead of translating English captions (ablation)object-mining captioner · Multi30k FrenchCIDEr: 0.099 -> 0.120Improved

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