AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Distractors-Immune Representation Learning with Cross-modal Contrastive Regularization for Change Captioning

Yunbin Tu, Liang Li, Li Su, Chenggang Yan, Qingming Huang

ECCV 2024 · Vision-language and multimodal › Cross-modal learning · approach: Contrastive vision-language alignment · 20 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 DIRL to vanilla Transformer (ablation)Transformer change captioner · CLEVR-DCCIDEr: 79.6 -> 81.8Improved
add CCR to vanilla Transformer (ablation)Transformer change captioner · CLEVR-DCCIDEr: 79.6 -> 82.7Improved
add both DIRL and CCR (ablation)Transformer change captioner · CLEVR-DCCIDEr: 79.6 -> 84.1Improved
add both DIRL and CCR (ablation)Transformer change captioner · CLEVR-DCSPICE: 15.7 -> 16.8Improved
use only one DIRL term (non-perturbation or discrimination) instead of both (ablation)DIRL · CLEVR-DCCIDEr: 79.6 -> 79.3 / 79.5No clear effect
combine both DIRL terms (correlate matching channels and decorrelate different ones) (ablation)DIRL · CLEVR-DCCIDEr: 79.3 / 79.5 -> 81.8Improved
DIRL+CCR vs prior SOTA SCORER+CBRDIRL+CCR · CLEVR-DCBLEU-4: 49.4 -> 51.4Improved
DIRL+CCR vs SCORER+CBR on METEORDIRL+CCR · CLEVR-DCMETEOR: 33.4 -> 32.3Hurt
DIRL+CCR vs VIXEN-CLIP pretrained on large editing dataDIRL+CCR · Image Editing RequestCIDEr: 38.1 -> 34.1Hurt
DIRL+CCR vs SMART on Spot-the-DiffDIRL+CCR · Spot-the-DiffSPICE: 19.0 -> 19.9Improved
DIRL+CCR vs SMART on CLEVR-ChangeDIRL+CCR · CLEVR-ChangeCIDEr: 123.6 -> 123.6No clear effect

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