AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Manipulation Intention Understanding for Zero-Shot Composed Image Retrieval

Yuanmin Tang, Jing Yu, Keke Gai, Gang Xiong, Gaopeng Gou, Meikang Qiu, Qi Wu

AAAI 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Attribute-based zero-shot learning · 1 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 intent-CC3M pseudo-manipulation texts to training (ablation)De-MINDS · CIRRRecall@1: 26.5 -> 30.0Improved
add chain-of-thought prompting for text generation (ablation)De-MINDS · CIRRRecall@1: 26.7 -> 30.0Improved
add similarity-guided filtering of generated texts (ablation)De-MINDS · CIRRRecall@1: 27.3 -> 30.0Improved
add the intention-understanding module (learnable queries) (ablation)De-MINDS · CIRRRecall@1: 25.9 -> 30.0Improved
add global feature to the query (ablation)De-MINDS · CIRRRecall@1: 27.7 -> 30.0Improved
add adaptive gating instead of direct summation (ablation)De-MINDS · CIRRRecall@1: 28.2 -> 30.0Improved
add distillation loss (ablation)De-MINDS · CIRRRecall@1: 26.7 -> 30.0Improved
use symmetric contrastive distillation instead of cosine distillation (ablation)De-MINDS · CIRRRecall@1: 28.4 -> 30.0Improved
use LLaVA+GPT-4o texts instead of Qwen-VL generated texts (ablation)De-MINDS · CIRRRecall@1: 28.8 -> 30.0Improved
use Long-CLIP instead of intention understanding (ablation)De-MINDS · CIRRRecall@1: 30.0 -> 26.3Hurt
train baseline Context-I2W with intent-CC3M captionsContext-I2W · CIRRRecall@1: 25.6 -> 26.5Improved
use De-MINDS instead of PrediCIRPrediCIR comparison · CIRRRecall@1: 27.2 -> 30.0Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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