Crowd Counting on ShanghaiTech Part B
6.2MAEGauNet (ResNet-50)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GauNet (ResNet-50)Backbone=ResNet-50, Venue=Ours2022.06 | 6.2 | 9.9 | — | |
| P2PNet2021.07 | 6.25 | 9.9 | — | |
| SGANet2019.11 | 6.3 | — | 10.6 | |
| Inception-v3status=modified2019.11 | 6.4 | — | 9.8 | |
| ADSCNetVenue=CVPR'202021.07 | 6.4 | 11.3 | — | |
| ADSCNetVenue=CVPR 202022.06 | 6.4 | 11.3 | — | |
| SPANet2019.11 | 6.5 | — | 9.9 | |
| SANet + SPANetVenue=ICCV'192021.07 | 6.5 | 9.9 | — | |
| SGANetCurriculum Loss=true2019.11 | 6.6 | — | 10.2 | |
| S-DCNetVenue=ICCV'192021.07 | 6.7 | 10.7 | — | |
| AMSNetVenue=ECCV'202021.07 | 6.7 | 10.2 | — | |
| AMSNetVenue=ECCV'202022.06 | 6.7 | 10.2 | — | |
| HCA-addApproach=Classification approach2023.10 | 6.7 | — | 11.4 | |
| HCA-mulApproach=Classification approach2023.10 | 6.8 | — | 11.4 | |
| HCA-dApproach=Classification approach2023.10 | 6.8 | — | 11.8 | |
| CLSApproach=Classification approach2023.10 | 7 | — | 11.8 | |
| AMRNetVenue=ECCV'202021.07 | 7.02 | 11 | — | |
| GauNet (SANet)Backbone=SANet, Venue=Ours2022.06 | 7.1 | 11.2 | — | |
| DC-regressionApproach=Local count regression2023.10 | 7.1 | — | 11 | |
| SDANetYear=2024, Supervision Type=FS, Parameters (M)=56.52025.12 | 7.1 | — | 12 | |
| DeepCount2019.08 | 7.2 | — | 11.3 | |
| GLFmethod_type=single-class crowd counting2022.01 | 7.3 | — | — | |
| GLossVenue=CVPR'212022.06 | 7.3 | 11.7 | — | |
| GLApproach=Density map regression2023.10 | 7.3 | — | 11.7 | |
| DM-CountVenue=NeurIPS'202021.07 | 7.4 | 11.8 | — | |
| DM-CountVenue=NeurIPS'202022.06 | 7.4 | 11.8 | — | |
| MNAApproach=Density map regression2023.10 | 7.4 | — | 11.3 | |
| OTApproach=Density map regression2023.10 | 7.4 | — | 11.8 | |
| SRRNetYear=2023, Supervision Type=FS, Parameters (M)=66.142025.12 | 7.4 | — | 13.6 | |
| TMTBVenue=Ours, Labeled Percentage=40%2025.03 | 7.5 | — | 12.9 | |
| ADCrowdNet(AMG-DME)Architecture=AMG-DME2018.11 | 7.6 | 13.9 | — | |
| GauNet (CSRNet)Backbone=CSRNet, Venue=Ours2022.06 | 7.6 | 12.7 | — | |
| SFCNYear=2021, Supervision Type=FS, Parameters (M)=38.62025.12 | 7.6 | — | 13 | |
| ADCrowdNet(AMG-attn-DME)Architecture=AMG-attn-DME2018.11 | 7.7 | 12.9 | — | |
| Bayesian+Venue=ICCV'192021.07 | 7.7 | 12.7 | — | |
| BLApproach=Density map regression2023.10 | 7.7 | — | 12.7 | |
| CANNet2019.11 | 7.8 | — | 12.2 | |
| CANVenue=CVPR'192021.07 | 7.8 | 12.2 | — | |
| SDANetVenue=AAAI'202021.07 | 7.8 | 10.2 | — | |
| MRC-CrowdVenue=T-CSVT'24, Labeled Percentage=40%2025.03 | 7.8 | — | 13.3 | |
| RANet2019.11 | 7.9 | — | 12.9 | |
| SALCrowdVenue=ACM MM'24, Labeled Percentage=40%2025.03 | 7.9 | — | 12.7 | |
| Wan et al.2019.11 | 8.1 | — | 13.6 | |
| RPNetmethod_type=single-class crowd counting2022.01 | 8.1 | — | — | |
| PaDNetApproach=Density map regression2023.10 | 8.1 | — | 12.2 | |
| OT-MVenue=CVPR'23, Labeled Percentage=40%2025.03 | 8.1 | — | 13.1 | |
| TCFormerYear=2025, Supervision Type=NF, Parameters (M)=5.522025.12 | 8.155 | — | 14.024 | |
| ADCrowdNet(AMG-bAttn-DME)Architecture=AMG-bAttn-DME2018.11 | 8.2 | 15.7 | — | |
| TEDNet2019.11 | 8.2 | — | 12.8 | |
| LMSNetYear=2024, Supervision Type=FS, Parameters (M)=0.732025.12 | 8.2 | — | 13.5 | |
| ANF2019.11 | 8.3 | — | 13.2 | |
| SANetOutput size=1:12019.09 | 8.4 | — | 13.6 | |
| SANet2019.11 | 8.4 | — | 13.6 | |
| SANet2019.08 | 8.4 | — | 13.6 | |
| SANetVenue=ECCV'182022.06 | 8.4 | 13.2 | — | |
| DRCNApproach=Density map regression2023.10 | 8.5 | — | 14.4 | |
| RepMobileNetYear=2024, Supervision Type=FS, Parameters (M)=3.412025.12 | 8.6 | — | 13.7 | |
| DADNet2019.11 | 8.8 | — | 13.5 | |
| RAQNetYear=2024, Supervision Type=FS, Parameters (M)=42.772025.12 | 9 | — | 15.4 | |
| PSDDN+2019.08 | 9.1 | — | 14.2 | |
| LMSFFNetYear=2024, Supervision Type=FS, Parameters (M)=2.882025.12 | 9.2 | — | 15.1 | |
| ADCrowdNet(DME)Architecture=DME2018.11 | 9.3 | 16.9 | — | |
| SCNetOutput size=1:12019.09 | 9.3 | — | 14.4 | |
| DLPTNetYear=2024, Supervision Type=FS, Parameters (M)=110.92025.12 | 9.3 | — | 15.6 | |
| DACountVenue=ACM MM'22, Labeled Percentage=40%2025.03 | 9.6 | — | 14.6 | |
| CUVenue=ICCV'23, Labeled Percentage=10%2025.03 | 9.7 | — | 17.7 | |
| SALCrowdVenue=ACM MM'24, Labeled Percentage=10%2025.03 | 9.7 | — | 17.5 | |
| Zhang et al.Year=2024, Supervision Type=NF2025.12 | 9.7 | — | 17.5 | |
| TMTBVenue=Ours, Labeled Percentage=10%2025.03 | 9.8 | — | 17.2 | |
| SAFECountmethod_type=few-shot counting2022.01 | 9.98 | — | — | |
| Liu et al.Output size=1:42019.09 | 10.1 | — | 18.8 | |
| MRCNetOutput size=1:12019.09 | 10.3 | — | 18.4 | |
| MRC-CrowdVenue=T-CSVT'24, Labeled Percentage=10%2025.03 | 10.3 | — | 18.2 | |
| PDDNetYear=2024, Supervision Type=FS, Parameters (M)=1.12025.12 | 10.3 | — | 17 | |
| ic-CNNOutput size=1:12019.09 | 10.4 | — | 16.7 | |
| CSRNet2018.11 | 10.6 | 16 | — | |
| CSRNetOutput size=1:82019.09 | 10.6 | — | 16 | |
| CSRNet2019.11 | 10.6 | — | 16 | |
| CSRNet2019.08 | 10.6 | — | 16 | |
| CSRNetmethod_type=single-class crowd counting2022.01 | 10.6 | — | — | |
| CSRNetVenue=CVPR'182022.06 | 10.6 | 16 | — | |
| CSRNetApproach=Density map regression2023.10 | 10.6 | — | 16 | |
| TMTBVenue=Ours, Labeled Percentage=5%2025.03 | 10.6 | — | 19.6 | |
| ic-CNN2019.08 | 10.7 | — | 16 | |
| RegressionApproach=Local count regression2023.10 | 10.7 | — | 19.5 | |
| OT-MVenue=CVPR'23, Labeled Percentage=10%2025.03 | 10.8 | — | 18.2 | |
| TinyCountYear=2024, Supervision Type=FS, Parameters (M)=0.062025.12 | 10.8 | — | 18.4 | |
| OT-MYear=2023, Supervision Type=NF2025.12 | 10.8 | — | 18.2 | |
| DACountYear=2022, Supervision Type=NF2025.12 | 10.9 | — | 19.1 | |
| DHMoEYear=2025, Supervision Type=FS, Parameters (M)=5.862025.12 | 11 | — | 19.6 | |
| DACountVenue=ACM MM'22, Labeled Percentage=10%2025.03 | 11.1 | — | 19.1 | |
| FSIMSource=SHA, Adaptation=true2026.03 | 11.1 | 19.3 | — | |
| MRC-CrowdVenue=T-CSVT'24, Labeled Percentage=5%2025.03 | 11.7 | — | 17.8 | |
| URCVenue=ICCV'212022.06 | 12 | 18.7 | — | |
| MRLYear=2023, Supervision Type=NF2025.12 | 12.1 | — | 19.7 | |
| Granular Ball Guided Stable Latent Domain DiscoverySource=SHA, Adaptation=false2026.03 | 12.3 | 21.6 | — | |
| C2MoTSource=SHA, Adaptation=true2026.03 | 12.4 | 21.1 | — | |
| DACountVenue=ACM MM'22, Labeled Percentage=5%2025.03 | 12.6 | — | 22.8 | |
| OT-MVenue=CVPR'23, Labeled Percentage=5%2025.03 | 12.6 | — | 21.5 | |
| DGCCSource=SHA, Adaptation=false2026.03 | 12.6 | 24.6 | — |