List your company

Image Retrieval on MS-COCO

100R@1HQA-VLAttack
46.75260.57674.488.224Apr 13, 2020Apr 13, 2021Apr 13, 2022Apr 14, 2023Apr 13, 2024Apr 13, 2025Apr 14, 2026

Evaluation results

MethodLinks
HQA-VLAttackSurrogate Model=CLIPViT, Victim Model=CLIPViT2026.04100——
HQA-VLAttackSurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.04100——
HQA-VLAttackSurrogate Model=ALBEF, Victim Model=ALBEF2026.0499.97——
HQA-VLAttackSurrogate Model=TCL, Victim Model=TCL2026.0499.97——
SGASurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0499.8——
SGASurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0499.79——
DRASurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0499.7——
DRASurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0499.7——
SGASurrogate Model=TCL, Victim Model=TCL2026.0499.15——
DRASurrogate Model=TCL, Victim Model=TCL2026.0499.15——
Co-AttackSurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0498.8——
Co-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0498.56——
SGASurrogate Model=ALBEF, Victim Model=ALBEF2026.0496.95——
DRASurrogate Model=ALBEF, Victim Model=ALBEF2026.0496.47——
Sep-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0492.49——
Co-AttackSurrogate Model=TCL, Victim Model=TCL2026.0491.39——
Sep-AttackSurrogate Model=ALBEF, Victim Model=ALBEF2026.0489.88——
HQA-VLAttackSurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0489.26——
Co-AttackSurrogate Model=ALBEF, Victim Model=ALBEF2026.0487.83——
HQA-VLAttackSurrogate Model=ALBEF, Victim Model=TCL2026.0487.81——
HQA-VLAttackSurrogate Model=TCL, Victim Model=ALBEF2026.0486.32——
PGDSurrogate Model=ALBEF, Victim Model=ALBEF2026.0486.3——
HQA-VLAttackSurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0484.56——
PGDSurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0484.2——
HQA-VLAttackSurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0481.64——
HQA-VLAttackSurrogate Model=ALBEF, Victim Model=CLIPViT2026.0480.2——
Sep-AttackSurrogate Model=TCL, Victim Model=TCL2026.0478.97——
HQA-VLAttackSurrogate Model=TCL, Victim Model=CLIPCNN2026.0478.22——
Sep-AttackSurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0477.94——
HQA-VLAttackSurrogate Model=TCL, Victim Model=CLIPViT2026.0476.22——
DRASurrogate Model=TCL, Victim Model=ALBEF2026.0475.86——
SGASurrogate Model=TCL, Victim Model=ALBEF2026.0473.3——
DRASurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0469.99——
PGDSurrogate Model=TCL, Victim Model=TCL2026.0469.53——
DRASurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0469.22——
DRASurrogate Model=ALBEF, Victim Model=TCL2026.0467.46——
DRASurrogate Model=ALBEF, Victim Model=CLIPViT2026.0467.43——
X-FMEvaluation protocol=Fine-Tune, Model size=base-size2023.016787.292.4
PGDSurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0466.85——
SGASurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0466.52——
X2-VLMEvaluation protocol=Fine-Tune, Model size=base-size2023.0166.287.192.2
HQA-VLAttackSurrogate Model=CLIPViT, Victim Model=ALBEF2026.0466.03——
SGASurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0465.83——
SGASurrogate Model=TCL, Victim Model=CLIPCNN2026.0465.7——
DRASurrogate Model=TCL, Victim Model=CLIPCNN2026.0465.7——
SGASurrogate Model=ALBEF, Victim Model=TCL2026.0465.38——
mPLUG-2Evaluation protocol=Fine-Tune, Model size=base-size2023.0165.386.992.4
SGASurrogate Model=ALBEF, Victim Model=CLIPViT2026.0465.25——
HQA-VLAttackSurrogate Model=CLIPViT, Victim Model=TCL2026.0465.25——
DRASurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0464.87——
SGASurrogate Model=TCL, Victim Model=CLIPViT2026.0463.99——
DRASurrogate Model=TCL, Victim Model=CLIPViT2026.0463.99——
X-VLMEvaluation protocol=Fine-Tune, Model size=base-size2023.0163.185.791.6
Co-AttackSurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0462.51——
Co-AttackSurrogate Model=TCL, Victim Model=CLIPCNN2026.0462.49——
HQA-VLAttackSurrogate Model=CLIPCNN, Victim Model=TCL2026.0462.49——
BERT-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPCNN2026.0462.17——
HQA-VLAttackSurrogate Model=CLIPCNN, Victim Model=ALBEF2026.0462.13——
FALCON#Images=4M, Evaluation Protocol=Fine-tuned2025.0561.583.690.2
BERT-AttackSurrogate Model=TCL, Victim Model=CLIPCNN2026.0461.26——
Sep-AttackSurrogate Model=TCL, Victim Model=CLIPCNN2026.0461.26——
MAFA#Images=4M, Evaluation Protocol=Fine-tuned2025.0561.283.990.2
X-FMEvaluation protocol=Zero-Shot, Model size=base-size2023.0161.184.590.6
SGASurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0460.77——
Co-AttackSurrogate Model=TCL, Victim Model=CLIPViT2026.0460.46——
Sep-AttackSurrogate Model=TCL, Victim Model=CLIPViT2026.0460.13——
ALIGN#Images=1.2B, Evaluation Protocol=Fine-tuned2025.0559.983.389.8
GRIT-VLP#Images=4M, Evaluation Protocol=Fine-tuned2025.0559.683.389.9
TCL#Images=4M, Evaluation Protocol=Fine-tuned2025.055983.289.9
Co-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0458.9——
BERT-AttackSurrogate Model=TCL, Victim Model=CLIPViT2026.0458.8——
OmniVLEvaluation protocol=Fine-Tune, Model size=base-size2023.0158.582.689.5
OSCARLModel scale=Large2020.0457.582.889.8
BERT-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0457.49——
Sep-AttackSurrogate Model=CLIPCNN, Victim Model=CLIPViT2026.0457.11——
Co-AttackSurrogate Model=TCL, Victim Model=ALBEF2026.0457.09——
ALBEF#Images=4M, Evaluation Protocol=Fine-tuned2025.0556.881.589.2
Sep-AttackSurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0455.88——
Co-AttackSurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0455.64——
X-VLMEvaluation protocol=Zero-Shot, Model size=base-size2023.0155.682.790
BERT-AttackSurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0455.32——
FALCON#Images=4M, Evaluation Protocol=Zero-shot2025.0554.879.887.6
BERT-AttackSurrogate Model=ALBEF, Victim Model=CLIPCNN2026.0454.75——
Co-AttackSurrogate Model=ALBEF, Victim Model=CLIPViT2026.0454.75——
Sep-AttackSurrogate Model=ALBEF, Victim Model=CLIPViT2026.0454.46——
Uni-Perceiver-MoEEvaluation protocol=Fine-Tune, Model size=base-size2023.0154.1——
OSCARBModel scale=Base2020.045480.888.5
MAFA#Images=4M, Evaluation Protocol=Zero-shot2025.0553.979.687.7
TCL#Images=4M, Evaluation Protocol=Zero-shot2025.0553.57987.1
Sep-AttackSurrogate Model=TCL, Victim Model=ALBEF2026.0452.97——
UNITER#Images=4M, Evaluation Protocol=Fine-tuned2025.0552.979.988
BERT-AttackSurrogate Model=ALBEF, Victim Model=CLIPViT2026.0452.28——
SoTALModel scale=Large2020.0451.778.486.9
BERT-AttackSurrogate Model=CLIPViT, Victim Model=CLIPViT2026.0451.68——
Uni-Perceiver-MoEEvaluation protocol=Zero-Shot, Model size=base-size2023.0151.6——
ImageBERT#Images=6M, Evaluation Protocol=Fine-tuned2025.0550.578.787.1
FLAVAEvaluation protocol=Fine-Tune, Model size=base-size2023.0150.174.483.2
ALBEF#Images=4M, Evaluation Protocol=Zero-shot2025.0550.176.484.5
Sep-AttackSurrogate Model=CLIPViT, Victim Model=CLIPCNN2026.0449.76——
UNITER#Images=4M, Evaluation Protocol=Zero-shot2025.0548.876.785.8