AI Paper Atlas

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

Adapting In-context Generation for Enhanced Composed Image Retrieval

Haiwen Li, Zining Chen, Delong Liu, Zhaohui Hou, Zhicheng Zhao, Fei Su

CVPR 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Composed and text-based image retrieval

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 DAIG synthetic pretraining + FRA to SPRC at 32-shotSPRC · CIRR testRecall@5: 57.61 -> 72.41Improved
add DAIG to BLIP4CIR at 32-shotBLIP4CIR · CIRR testRecall@5: 27.86 -> 57.01Improved
add DAIG to SPRC with full labelsSPRC · CIRR testRecall@5: 82.22 -> 84.10Improved
add DAIG to SPRC with full labelsSPRC · FashionIQ valAverage Recall@50: 74.95 -> 75.96Improved
add DAIG to CLIP4CIR at 1% labels on FashionIQ DressCLIP4CIR · FashionIQ val DressRecall@10: 24.79 -> 24.74No clear effect
use DAIG triplets vs PTG for few-shot SPRC (8-shot)SPRC · FashionIQAvg@10: 30.6 -> 43.7Improved
train SPRC on 20k DAIG triplets instead of 534k CIRHS (ablation)SPRC · CIRRAvg.: 69.14 -> 71.68Improved
use domain-adapted LoRA T2I instead of zero-shot in-context generation (ablation)SPRC · CIRRRecall@5: 66.10 -> 70.19Improved
replace LoRA with CIR-LoRA (MoE) (ablation)SPRC · CIRRRecall@5: 70.19 -> 71.01Improved
add visual feature perturbation (p=0.5) in DRSP (ablation)SPRC · CIRRRecall@5: 71.01 -> 72.02Improved
add angular margin in FRA (ablation)SPRC · CIRRRecall@5: 74.65 -> 75.52Improved
increase number of synthetic triplets beyond 30k (ablation)SPRC + DAIG · CIRR/FashionIQRecall@5No 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