HyperAIHyperAI超神経

Command Palette

Search for a command to run...

プラットフォーム

Zero Shot Composed Image Retrieval Zs Cir On 2

評価指標

(Recall@10+Recall@50)/2

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

Paper Title
RTD + LinCIR (CLIP G/14)56.74An Efficient Post-hoc Framework for Reducing Task Discrepancy of Text Encoders for Composed Image Retrieval
LinCIR (CLIP G/14)55.40Language-only Efficient Training of Zero-shot Composed Image Retrieval
SEIZE (CLIP G/14)54.45Semantic Editing Increment Benefits Zero-Shot Composed Image Retrieval
CoLLM (finetuned - BLIP-L/16)49.9CoLLM: A Large Language Model for Composed Image Retrieval
CoVR-BLIP-248.3CoVR-2: Automatic Data Construction for Composed Video Retrieval
MagicLens (CoCa L)48.1MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions
OSrCIR (CLIP G/14)47.34Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval
WeiMoCIR (CLIP G/14)47.16Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity
MTCIR (CLIP L/14)46.42Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval
CompoDiff (CLIP G/14)45.37CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion
MagicLens (CoCa B)45.3MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions
CoLLM (Pretrained - BLIP-L/16)45.3CoLLM: A Large Language Model for Composed Image Retrieval
TransAgg (Laion-CIR-Combined)44.75Zero-shot Composed Text-Image Retrieval
WeiMoCIR (CLIP H/14)44.58Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity
CompoDiff (CLIP L/14)44.11CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion
LDRE (CLIP G/14)43.98LDRE: LLM-based Divergent Reasoning and Ensemble for Zero-Shot Composed Image Retrieval
OSrCIR (CLIP B/32)42.87Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval
OSrCIR (CLIP L/14)42.82Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval
CIReVL (CLIP G/14)42.28Vision-by-Language for Training-Free Compositional Image Retrieval
MagicLens (CLIP L)41.6MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions
0 of 40 row(s) selected.
Zero Shot Composed Image Retrieval Zs Cir On 2 | SOTA | HyperAI超神経