List your company

Object Counting on FSC-147 (val)

5.71MAEABACUS-3B
5.30888.016910.72513.4331Jan 22, 2022Oct 17, 2022Jul 13, 2023Apr 7, 2024Dec 31, 2024Sep 26, 2025Jun 22, 2026

Evaluation results

MethodLinks
ABACUS-3BSupervision=Zero-shot/MLLM (Z)2026.065.7126.46—
CountGD2026.027.126.08—
CountGDClass Spec.=VE & Text2025.117.126.08—
CountGDScheme=Few-shot, Exemplar=Visual & Text, Backbone=GDINO-Swin-B2026.037.126.08—
CountGDText prompt=Yes, Box exemplar=Yes2026.057.126.08—
CountOCCClass Spec.=VE & Text2025.117.2427.1—
CountSEScheme=Few-shot, Exemplar=Text, Backbone=GDINO-Swin-B2026.038.5154.93—
DAVEFew-shot=true2024.048.9128.08—
DAVE2026.028.9128.08—
DAVEClass Spec.=Visual Exemplars2025.118.9128.08—
DAVEText prompt=No, Box exemplar=Yes2026.058.9128.08—
CoDiProtocol=One-shot, Category=Localization-based, Text-prompts=false2025.128.9235.89—
CAD-GD2026.029.340.96—
CAD-GDType=Text, Base model=Grounding DINO, Differentiable=×2026.069.340.96—
CountDiffProtocol=One-shot, Category=Density-based, Text-prompts=true2025.129.3533.44—
GeCoText prompt=No, Box exemplar=Yes2026.059.5243—
TrueCount𝑡𝑥𝑡Type=Detection, Backbone=GDino, Decoder=DefTransformer2026.069.5448.01—
CountEx2026.029.5938.54—
DCountType=Detection, Backbone=GDino, Decoder=DefTransformer2026.069.6353.6—
SSDBackbone=ResNet-50, Shot count=3-shot2024.059.7329.72—
DAVEboxCounting mode=detection-based2024.049.7540.3—
GeComode=one-shot2024.099.9737.85—
GeCoProtocol=One-shot, Category=Detection-based, Text-prompts=false2025.129.9737.85—
GroundingRECSupervision=point-level2026.0210.0658.62—
GroundingREC2026.0210.0658.62—
GroundingRECClass Spec.=Text2025.1110.0658.62—
GroundingRECType=Text, Base model=Grounding DINO, Differentiable=×2026.0610.0658.62—
GroundingRECType=Detection, Backbone=GDino, Decoder=DefTransformer2026.0610.0658.62—
LOCAScenario=few-shot2022.1110.2432.56—
LOCA2023.0710.2432.56—
LOCAFew-shot=true2024.0410.2432.56—
LOCABackbone=ResNet-50, Shot count=3-shot2024.0510.2432.56—
LOCAScheme=Few-shot, Shot=32025.0210.2432.56—
LOCA2026.0210.2432.56—
LOCAClass Spec.=Visual Exemplars2025.1110.2432.56—
LOCAScheme=Few-shot, Exemplar=Visual, Backbone=ResNet-502026.0310.2432.56—
LoCAText prompt=No, Box exemplar=Yes2026.0510.2432.56—
CACVITBackbone=ViT-B, Resolution=384, Shots=32023.0510.6337.95—
CACViT2026.0210.6337.95—
HieraCountText prompt=Yes, Box exemplar=Yes2026.0510.8349.36—
DAVEmode=one-shot2024.0910.9843.26—
DAVEProtocol=One-shot, Category=Detection-based, Text-prompts=false2025.1210.9843.26—
SSDBackbone=ResNet-50, Shot count=1-shot2024.0511.0334.83—
LOCAEvaluation Protocol=one-shot2022.1111.3638.04—
LOCABackbone=ResNet-50, Shot count=1-shot2024.0511.3638.04—
LOCAmode=one-shot2024.0911.3638.04—
LOCAProtocol=One-shot, Category=Density-based, Text-prompts=false2025.1211.3638.04—
CACVITBackbone=ViT-B, Resolution=384, Shots=12023.0511.4141.04—
CountGDSupervision=point-level2026.0212.1447.51—
CountGDtxtClass Spec.=Text2025.1112.1447.51—
CountGDScheme=Zero-shot, Exemplar=Text, Backbone=GDINO-Swin-B2026.0312.1447.51—
CountGDType=Text, Base model=Grounding DINO, Differentiable=×2026.0612.1447.51—
COUNTGDType=Detection, Backbone=GDino, Decoder=DefTransformer2026.0612.1447.51—
CountGD++Supervision=Point-level (P)2026.0612.1447.51—
RT-CounterType=Text, Base model=YOLOE, Differentiable=✓2026.0612.5651.25—
QICA†Scheme=Zero-shot, Exemplar=Text, Backbone=ViT-L/142026.0312.9856.35—
CounTRYear=2022, Backbone=ViT, # Shots=32022.0813.1349.83—
CounTRScenario=few-shot2022.1113.1349.83—
CounTRBackbone=ViT-B & CNN, Resolution=384, Shots=32023.0513.1349.83—
CounTR2023.0713.1349.83—
CounTRFew-shot=true2024.0413.1349.83—
CounTRBackbone=ViT/ConvNet, Shot count=3-shot2024.0513.1349.83—
CounTR2026.0213.1349.83—
CounTRClass Spec.=Visual Exemplars2025.1113.1349.83—
CountTRScheme=Few-shot, Exemplar=Visual, Backbone=ViT-B/162026.0313.1349.83—
CounTREvaluation Protocol=one-shot2022.1113.1549.72—
CounTRBackbone=ViT-B & CNN, Resolution=384, Shots=12023.0513.1549.72—
CounTRBackbone=ViT/ConvNet, Shot count=1-shot2024.0513.1549.72—
CounTRmode=one-shot2024.0913.1549.72—
CounTRProtocol=One-shot, Category=Density-based, Text-prompts=false2025.1213.1549.72—
MambaCountType=Point, Backbone=YOLOE, Decoder=Mamba2026.0613.2253.62—
T2ICountScheme=Zero-shot, Shot=02025.0213.7858.78—
T2ICountSupervision=point-level2026.0213.7858.78—
T2ICountScheme=Zero-shot, Exemplar=Text, Backbone=Stable Diff-v1.52026.0313.7858.78—
T2ICountSupervision=Point-level (P)2026.0613.7858.78—
QICAScheme=Zero-shot, Exemplar=Text, Backbone=ViT-B/162026.0313.8260.24—
SPDCNBackbone=ResNet-18, Resolution=576, Shots=32023.0514.5949.97—
SPDCNBackbone=VGG-19, Shot count=3-shot2024.0514.5949.97—
WS-COCSupervision=image-level2026.0214.7754.24—
WS-COC-7BSupervision=Image-level (I)2026.0614.7754.24—
YOLO-CountScheme=Zero-shot, Exemplar=Text, Backbone=YOLOv8-L2026.0314.896.14—
GeCoScheme=Zero-shot, Shot=02025.0214.8164.95—
GeCoScheme=Zero-shot, Exemplar=Visual, Backbone=SAM-ViT-H/162026.0314.8164.95—
SAFECounttest-time adaptation=false2022.0115.2847.2—
SAFECountScenario=few-shot2022.1115.2847.2—
SAFECountBackbone=ResNet-18, Resolution=512, Shots=32023.0515.2847.2—
SAFECountFew-shot=true2024.0415.2847.2—
SAFECountBackbone=ResNet-18, Shot count=3-shot2024.0515.2847.2—
PseCoScheme=Few-shot, Shot=32025.0215.3168.34—
PseCoScheme=Few-shot, Exemplar=Visual, Backbone=SAM-ViT-H/162026.0315.3168.34—
DAVEprompt-based=true2024.0415.4852.57—
DAVEScheme=Zero-shot, Shot=02025.0215.4852.57—
DAVEScheme=Zero-shot, Exemplar=Visual, Backbone=ResNet-502026.0315.4852.57—
DAVEprmType=Text, Base model=ResNet-50, Differentiable=×2026.0615.4852.57—
CountDiffSupervision=point-level2026.0215.554.33—
DAVE0-shotsetup=zero-shot2024.0415.5452.67—
DAVECategory=CAC, Backbone=ResNet-50, Resolution=5122024.0915.5452.67—
DAVEType=Zero-shot visual prototype, Base model=ResNet-50, Differentiable=×2026.0615.5452.67—
BMNet+Fine-tuning=false, Backbone=SwAV2022.0315.74—58.53
BMNet+reference_type=reference-based, shot=3-shot2022.0515.7458.53—