List your company
ResearchFormerly WizWand

Visual Grounding on Who's Waldo (test)

63.5AccuracyWho's Waldo
22.52433.16243.854.438Aug 16, 2021Jan 2, 2022May 21, 2022Oct 7, 2022Feb 23, 2023Jul 12, 2023Nov 28, 2023
Updated 8mo ago

Evaluation results

MethodLinks
Who's WaldoTraining Paradigm=Supervised2023.1163.5
Who's Waldo modelTraining Data=Who's Waldo2021.0863.5
VR-VLA (Ours+VR-CLIP)Foundation Model=CLIP, Variant=VR-CLIP2023.1161.3
VR-VLA (Ours)Foundation Model=CLIP2023.1160.8
VR-VLA (Ours+VR-FLAVA)Foundation Model=FLAVA, Variant=VR-FLAVA2023.1159.8
VR-VLA (Ours)Foundation Model=FLAVA2023.1159.6
L→R (Largest)Type=Baseline2021.0857.7
Big->SmallType=Baseline2021.0848.2
SL-CCRFTraining Paradigm=Pretrained on grounding data2023.1146.4
SL-CCRFTraining Data=Flickr30K Entities, Name processing configuration=Constant2021.0846.4
SL-CCRFTraining Data=Flickr30K Entities, Name processing configuration=Random2021.0844.1
MAttNetTraining Data=RefCOCOg, Name processing configuration=Random2021.0844
MAttNetTraining Data=RefCOCOg, Name processing configuration=Full Names2021.0843.6
SL-CCRFTraining Data=Flickr30K Entities, Name processing configuration=Full Names2021.0843.5
Gupta et al.Training Data=Flickr30K Entities, Name processing configuration=Random2021.0841.1
Gupta et al.Training Data=Flickr30K Entities, Name processing configuration=Full Names2021.0839.3
Gupta et al.Training Data=COCO, Name processing configuration=Random2021.0839.3
UNITERTraining Data=Multiple [36, 30, 44, 56], Name processing configuration=Random2021.0838.4
L→R (All)Type=Baseline2021.0838.4
Gupta et al. (Flickr30K)Training Paradigm=Pretrained on grounding data, Pre-training Dataset=Flickr30K2023.1138.2
Gupta et al.Training Data=Flickr30K Entities, Name processing configuration=Constant2021.0838.2
Gupta et al.Training Data=COCO, Name processing configuration=Full Names2021.0836.9
UNITERTraining Data=Multiple [36, 30, 44, 56], Name processing configuration=Full Names2021.0836.3
Gupta et al. (COCO)Training Paradigm=Pretrained on grounding data, Pre-training Dataset=COCO2023.1135.6
Gupta et al.Training Data=COCO, Name processing configuration=Constant2021.0835.6
UNITERTraining Paradigm=Pretrained on grounding data2023.1134.2
UNITERTraining Data=Multiple [36, 30, 44, 56], Name processing configuration=Constant2021.0834.2
RandomType=Baseline2021.0830.9
ReCLIPFoundation Model=CLIP2023.1129.4
MAttNetTraining Paradigm=Pretrained on grounding data2023.1124.1
MAttNetTraining Data=RefCOCOg, Name processing configuration=Constant2021.0824.1