AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Aligning Information Capacity Between Vision and Language via Dense-to-Sparse Feature Distillation for Image-Text Matching

Yang Liu, Wentao Feng, Zhuoyao Liu, Shudong Huang, Jiancheng Lv

ICCV 2025 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 7 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 dense-to-sparse distillation to pretrain+align fine-tuning (ablation)D2S-VSE · Flickr30K 1K testrSum: 525.7 -> 531.9Improved
add dense-text pretraining before sparse fine-tuning (ablation)D2S-VSE · Flickr30K 1K testrSum: 509.5 -> 525.7Improved
add sparse-text alignment loss alongside pretrain and distillation (ablation)D2S-VSE · Flickr30K 1K testrSum: 512.1 -> 531.9Improved
fine-tune on sparse text after dense pretraining (ablation)D2S-VSE · Flickr30K 1K testrSum: 425.2 -> 525.7Improved
add distillation without sparse alignment on top of pretraining only (ablation)D2S-VSE · Flickr30K 1K testrSum: 425.2 -> 512.1Improved
use negative cosine similarity rather than l1 distance for distillation (ablation)D2S-VSE · Flickr30K 1K testrSum: 528.9 -> 531.9Improved
use negative cosine similarity rather than l2 distance for distillation (ablation)D2S-VSE · Flickr30K 1K testrSum: 530.2 -> 531.9Improved
place learnable tokens surrounding the sparse text rather than as prefix (ablation)D2S-VSE · Flickr30K 1K testrSum: 529.3 -> 531.9Improved
place learnable tokens surrounding the sparse text rather than as postfix (ablation)D2S-VSE · Flickr30K 1K testrSum: 529.5 -> 531.9Improved
increase the number of learnable tokens from 100 to 120 (ablation)D2S-VSE · Flickr30K 1K testR@1No clear effect
increase decoder attention heads beyond 4 (ablation)D2S-VSE · Flickr30K 1K testR@1: within 0.2%No clear effect
increase decoder depth beyond 4 layers (ablation)D2S-VSE · Flickr30K 1K testR@1Hurt

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