Counting on CARPK
1.99MAEAdaCount
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AdaCountTraining=false2026.07 | 1.99 | 3.05 | |
| CountGD++Training=true2026.07 | 2.48 | 3.74 | |
| SAM3CountTraining=false2026.07 | 3.11 | 5.6 | |
| SAM3Training=false2026.07 | 3.52 | 6.57 | |
| CountGDExemplar=Visual & Text, Fine-tuned=false2026.03 | 3.68 | 5.17 | |
| CountGDInference-only=false2025.12 | 3.83 | 5.41 | |
| CountGD2026.06 | 3.83 | 5.41 | |
| CountGD2026.06 | 3.83 | 5.41 | |
| CountGDTraining=true2026.07 | 3.83 | 5.41 | |
| MambaCount2026.06 | 4.31 | 7.42 | |
| RT-Counter2026.06 | 4.81 | 7.61 | |
| CACViTFine-tuned=true2023.05 | 4.91 | 6.49 | |
| SAFECountTraining Protocol=Fine-tuned on CARPK2022.01 | 5.33 | 7.04 | |
| SAFECountFine-tuned=true2023.05 | 5.33 | 7.04 | |
| CounTRFine-tuned=true2023.05 | 5.75 | 7.45 | |
| CountTRExemplar=Visual, Fine-tuned=true2026.03 | 5.75 | 7.45 | |
| BMNet+Fine-tuned=true2023.05 | 5.76 | 7.83 | |
| CountGDExemplar=Text, Fine-tuned=false2026.03 | 5.98 | 7.41 | |
| QICA†Exemplar=Text, Fine-tuned=false, Backbone=ViT-L/142026.03 | 6.07 | 7.82 | |
| QICATraining=true2026.07 | 6.07 | 7.82 | |
| CounTX2026.06 | 6.13 | 10.87 | |
| CountTX2026.06 | 6.13 | 10.87 | |
| QICAExemplar=Text, Fine-tuned=false, Backbone=ViT-B/162026.03 | 6.38 | 8.2 | |
| VLCounterLearnable Parameters (M)=1.44, MACs (G)=34.362023.12 | 6.46 | 8.68 | |
| VLCounterInference-only=false2025.12 | 6.46 | 8.68 | |
| VLCounterSupervision=point-level2026.02 | 6.46 | 8.68 | |
| VLCounterExemplar=Text, Fine-tuned=false2026.03 | 6.46 | 8.68 | |
| VLCounter2026.06 | 6.46 | 8.68 | |
| VLCounter2026.06 | 6.46 | 8.68 | |
| VLCounterTraining=true2026.07 | 6.46 | 8.68 | |
| CountZESInference-only=true2025.12 | 7.24 | 9.22 | |
| GMNTraining Protocol=Fine-tuned on CARPK2022.01 | 7.48 | 9.9 | |
| CounTXInference-only=false2025.12 | 8.13 | 10.87 | |
| CounTXExemplar=Text, Fine-tuned=false2026.03 | 8.13 | 10.87 | |
| CACViTFine-tuned=false2023.05 | 8.3 | 11.18 | |
| T2ICountInference-only=false2025.12 | 8.61 | 13.47 | |
| T2ICountSupervision=point-level2026.02 | 8.61 | 13.47 | |
| T2ICountExemplar=Text, Fine-tuned=false2026.03 | 8.61 | 13.47 | |
| T2ICountTraining=true2026.07 | 8.61 | 13.47 | |
| LOCAExemplar=Visual, Fine-tuned=false2026.03 | 9.97 | 12.51 | |
| VLPGSupervision=point-level2026.02 | 10.14 | 13.79 | |
| CountDiffSupervision=point-level2026.02 | 10.32 | 12.92 | |
| GeCoInference-only=false2025.12 | 10.34 | 14.73 | |
| GeCoTraining=true2026.07 | 10.34 | 14.73 | |
| WS-COCSupervision=image-level2026.02 | 10.39 | 15.83 | |
| BMNet+Fine-tuned=false2023.05 | 10.44 | 13.77 | |
| VA-CountInference-only=false2025.12 | 10.63 | 13.2 | |
| VA-CountExemplar=Text, Fine-tuned=false2026.03 | 10.63 | 13.2 | |
| CLIP-CountLearnable Parameters (M)=16.36, MACs (G)=123.062023.12 | 11.7 | 13.94 | |
| CLIP-CountInference-only=false2025.12 | 11.96 | 16.61 | |
| CLIP-CountSupervision=point-level2026.02 | 11.96 | 16.61 | |
| CLIP-CountExemplar=Text, Fine-tuned=false2026.03 | 11.96 | 16.61 | |
| CLIP-count2026.06 | 11.96 | 16.61 | |
| CLIP-count2026.06 | 11.96 | 16.61 | |
| OmniCountInference-only=true2025.12 | 13.41 | 16.85 | |
| OmniCountTraining=false2026.07 | 13.41 | 16.85 | |
| RCACFine-tuned=true2023.05 | 13.62 | 19.08 | |
| WS-COC-BaseSupervision=image-level2026.02 | 14.06 | 27.6 | |
| TFOCInference-only=true2025.12 | 14.35 | 17.22 | |
| TFOCTraining=false2026.07 | 14.35 | 17.22 | |
| SAFECountTraining Protocol=Pre-trained on FSC-1472022.01 | 16.66 | 24.08 | |
| SAFECountFine-tuned=false2023.05 | 16.66 | 24.08 | |
| RCACFine-tuned=false2023.05 | 17.98 | 24.21 | |
| FamNetTraining Protocol=Fine-tuned on CARPK2022.01 | 18.19 | 33.66 | |
| FamNet+Fine-tuned=true2023.05 | 18.19 | 33.66 | |
| T2ICount*Supervision=image-level2026.02 | 19.42 | 23.48 | |
| CountingDINOInference-only=true2025.12 | 21.26 | 28.2 | |
| CountingDINOTraining=false2026.07 | 21.26 | 28.2 | |
| CLIP-Count*Supervision=image-level2026.02 | 24.9 | 29.12 | |
| MLLM-ZeroSupervision=none2026.02 | 26.24 | 55.11 | |
| FamNetTraining Protocol=Pre-trained on FSC-1472022.01 | 28.84 | 44.47 | |
| FamNet+Fine-tuned=false2023.05 | 28.84 | 44.47 | |
| G-DINOInference-only=false2025.12 | 29.72 | 31.6 | |
| GMNTraining Protocol=Pre-trained on FSC-1472022.01 | 32.92 | 39.88 |