Object Detection on Watercolor2k (test)
89.81mAP (Overall)XL Model w/ All Images
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| XL Model w/ All ImagesSupervision Type=FS, Training Images=All, Model Architecture=XL2022.11 | 89.81 | — | — | — | — | — | — | — | — | |
| Train w/ All ImagesSupervision Type=FS, Training Images=All2022.11 | 89.17 | — | — | — | — | — | — | — | — | |
| Train w/ 512 ImagesSupervision Type=FS, Training Images=5122022.11 | 77.63 | — | — | — | — | — | — | — | — | |
| Inoue et al.Supervision Type=FS2022.11 | 77.3 | — | — | — | — | — | — | — | — | |
| Train w/ 256 ImagesSupervision Type=FS, Training Images=2562022.11 | 73.83 | — | — | — | — | — | — | — | — | |
| H2FA R-CNNSupervision Type=WS2022.11 | 73.8 | — | — | — | — | — | — | — | — | |
| Inoue et al.Supervision Type=WS2022.11 | 73.2 | — | — | — | — | — | — | — | — | |
| Teacher-StudentSupervision Type=SS2022.11 | 71.53 | — | — | — | — | — | — | — | — | |
| DASS-DetectorSetting=Unsupervised domain adaptation object detection, extra_data=true2023.09 | 71.5 | — | — | — | — | — | — | — | — | |
| Train w/ 64 ImagesSupervision Type=FS, Training Images=642022.11 | 71.24 | — | — | — | — | — | — | — | — | |
| UMTSupervision Type=SS2022.11 | 69.9 | — | — | — | — | — | — | — | — | |
| D-adaptSupervision Type=SS2022.11 | 68.9 | — | — | — | — | — | — | — | — | |
| All StylesSupervision Type=NS2022.11 | 67.36 | — | — | — | — | — | — | — | — | |
| MILABackbone=ResNet-101, Source Domain=PASCAL VOC2023.09 | 63.7 | — | 59 | 58.3 | 40.6 | 47.8 | 79.3 | — | 97.4 | |
| MILABackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 63.7 | — | 59 | 58.3 | 40.6 | 47.8 | 79.3 | — | 97.4 | |
| MILABackbone=ResNet-1012023.09 | 63.7 | — | 59 | 58.3 | 40.6 | 47.8 | 79.3 | — | 97.4 | |
| H2FA R-CNNSetting=Cross domain weakly supervised object detection2023.09 | 59.9 | — | — | — | — | — | — | — | — | |
| ATBackbone=ResNet-1012021.11 | 59.9 | — | 56.1 | 58.9 | 37.3 | 39.6 | 73.8 | — | 93.6 | |
| AT†Backbone=ResNet-101, Source Domain=PASCAL VOC, Note=Results copied from origin paper [17]2023.09 | 59.9 | — | 56.1 | 58.9 | 37.3 | 39.6 | 73.8 | — | 93.6 | |
| ATBackbone=ResNet-101, Note=results copied from the origin paper [17]2023.09 | 59.9 | — | 56.1 | 58.9 | 37.3 | 39.6 | 73.8 | — | 93.6 | |
| ATBackbone=ResNet-101, Source Domain=PASCAL VOC2023.09 | 59.7 | — | 57.2 | 57.2 | 34.2 | 36.9 | 78.5 | — | 94.3 | |
| ATBackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 59.7 | — | 57.2 | 57.2 | 34.2 | 36.9 | 78.5 | — | 94.3 | |
| ATBackbone=ResNet-1012023.09 | 59.7 | — | 57.2 | 57.2 | 34.2 | 36.9 | 78.5 | — | 94.3 | |
| Pseudo-Labelingbaseline detector=SSD300, extra annotations=true2018.03 | 59.1 | 84.8 | 57.7 | 48 | 44.9 | 46.6 | 72.6 | — | — | |
| DT+PLbaseline detector=SSD300, extra annotations=true2018.03 | 59.1 | 86.3 | 57.3 | 48.5 | 43 | 46.5 | 73.2 | — | — | |
| Oracle2018.12 | 58.6 | 83.6 | 59.4 | 50.7 | 43.7 | 39.5 | 74.5 | — | — | |
| Train-on-TargetMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 58.6 | 83.6 | 59.4 | 50.7 | 43.7 | 39.5 | 74.5 | — | — | |
| Ideal casebaseline detector=SSD3002018.03 | 58.4 | 76 | 60 | 52.7 | 41 | 43.8 | 77.3 | — | — | |
| UMTBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN, Cross-domain distillation=true, Target-like augmentation=true, Sample selection=true2020.03 | 58.1 | 88.2 | 55.3 | 51.7 | 39.8 | 43.6 | 69.9 | — | — | |
| UMTSetting=Unsupervised domain adaptation object detection2023.09 | 58.1 | — | — | — | — | — | — | — | — | |
| UMTBackbone=ResNet-1012021.11 | 58.1 | — | 55.3 | 51.7 | 39.8 | 43.6 | 69.9 | — | 88.2 | |
| UMTBackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 58.1 | — | 55.3 | 51.7 | 39.8 | 43.6 | 69.9 | — | 88.2 | |
| ICCMSetting=Cross domain weakly supervised object detection2023.09 | 57.4 | — | — | — | — | — | — | — | — | |
| Strong-Weak Domain AlignmentGlobal Alignment (G)=true, Instance-level Alignment (I)=true, Context-level Alignment (CTX)=true, Local Alignment (L)=true, CycleGAN (P)=true2018.12 | 56.7 | 90.5 | 54.8 | 49.4 | 38.6 | 38.8 | 67.9 | — | — | |
| PLGESetting=Cross domain weakly supervised object detection2023.09 | 56.5 | — | — | — | — | — | — | — | — | |
| MCARMultilabel-Conditional adversary=true, Prediction based Regularization=true2020.03 | 56 | 87.9 | 52.1 | 51.8 | 41.6 | 33.8 | 68.8 | — | — | |
| MCARSetting=Unsupervised domain adaptation object detection2023.09 | 56 | — | — | — | — | — | — | — | — | |
| UMTSCABackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN, Cross-domain distillation=true, Target-like augmentation=true2020.03 | 55.6 | 86.6 | 51.3 | 52.6 | 42.1 | 33.5 | 67.5 | — | — | |
| SCLMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 55.2 | 82.2 | 55.1 | 51.8 | 39.6 | 38.4 | 64 | — | — | |
| SCLBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN2020.03 | 55.2 | 82.2 | 55.1 | 51.8 | 39.6 | 38.4 | 64 | — | — | |
| SCLBackbone=ResNet-1012021.11 | 55.2 | — | 55.1 | 51.8 | 39.6 | 38.4 | 64 | — | 82.2 | |
| SCLBackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 55.2 | — | 55.1 | 51.8 | 39.6 | 38.4 | 64 | — | 82.2 | |
| MCARMultilabel-Conditional adversary=true, Prediction based Regularization=false2020.03 | 54.4 | 92.5 | 52.2 | 43.9 | 46.5 | 28.8 | 62.5 | — | — | |
| DT+PLbaseline detector=SSD3002018.03 | 54.3 | 76.5 | 54.9 | 46 | 37.4 | 38.5 | 72.3 | — | — | |
| DT+PLNet=SSD2018.10 | 54.3 | 76.5 | 54.9 | 46 | 37.4 | 38.5 | 72.3 | — | — | |
| UMTSCBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN, Cross-domain distillation=true, Target-like augmentation=false2020.03 | 54.2 | 79.7 | 49.5 | 50.1 | 45.5 | 30.6 | 69.8 | — | — | |
| Pseudo-Labelingbaseline detector=SSD3002018.03 | 54 | 76.3 | 54.9 | 46.6 | 37.5 | 36.9 | 71.7 | — | — | |
| UMTSBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN, Cross-domain distillation=false, Target-like augmentation=false2020.03 | 53.6 | 76.2 | 53.4 | 46.2 | 39.3 | 34.9 | 71.5 | — | — | |
| Strong-Weak Domain AlignmentGlobal Alignment (G)=true, Instance-level Alignment (I)=true, Context-level Alignment (CTX)=true, Local Alignment (L)=true2018.12 | 53.3 | 82.3 | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | — | |
| Strong-Weak DAG=true, I=true, CTX=true, L=true2018.12 | 53.3 | 82.3 | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | — | |
| SW-DAMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 53.3 | 82.3 | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | — | |
| SWDABackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN2020.03 | 53.3 | 82.3 | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | — | |
| SWDABackbone=ResNet-1012021.11 | 53.3 | — | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | 82.3 | |
| SWADABackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 53.3 | — | 55.9 | 46.5 | 32.7 | 35.5 | 66.7 | — | 82.3 | |
| Strong-Weak Domain AlignmentGlobal Alignment (G)=true, Instance-level Alignment (I)=true, Context-level Alignment (CTX)=true2018.12 | 52.1 | 79.4 | 54.8 | 47.2 | 37.1 | 31.5 | 62.4 | — | — | |
| OursSource Domain=PASCAL VOC, n=32019.05 | 52 | — | — | — | — | — | — | — | — | |
| DMBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN2020.03 | 52 | — | — | — | — | — | — | — | — | |
| I³NetAblation=Full2021.03 | 51.5 | 81.1 | 49.3 | 46.2 | 35 | 31.9 | 65.7 | — | — | |
| Faster RCNN*CycleGAN (P)=true2018.12 | 51.2 | 83.3 | 52.7 | 45.3 | 33.1 | 28.8 | 64 | — | — | |
| I³NetAblation=w/o COPM2021.03 | 50.9 | 75.6 | 49.2 | 45.9 | 37.9 | 33.2 | 63.6 | — | — | |
| OracleBackbone=ResNet-1012021.11 | 50.6 | — | 49.7 | 42.5 | 38.7 | 52.1 | 68.6 | — | 51.8 | |
| OracleBackbone=ResNet-101, Source Domain=PASCAL VOC2023.09 | 50.6 | — | 49.7 | 42.5 | 38.7 | 52.1 | 68.6 | — | 51.8 | |
| OracleBackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 50.6 | — | 49.7 | 42.5 | 38.7 | 52.1 | 68.6 | — | 51.8 | |
| OracleBackbone=ResNet-1012023.09 | 50.6 | — | 49.7 | 42.5 | 38.7 | 52.1 | 68.6 | — | 51.8 | |
| I³NetAblation=w/o DCBR2021.03 | 50.5 | 78.7 | 49.2 | 42.6 | 37.4 | 32.4 | 62.5 | — | — | |
| Domain Transferbaseline detector=SSD3002018.03 | 50.4 | 82.8 | 47 | 40.2 | 34.6 | 35.3 | 62.5 | — | — | |
| Strong-Weak Domain AlignmentGlobal Alignment (G)=true, Instance-level Alignment (I)=true, Local Alignment (L)=true2018.12 | 50.4 | 71.3 | 52 | 46.6 | 36.2 | 29.2 | 67.3 | — | — | |
| Strong-Weak DAG=true, I=true, CTX=true, L=false2018.12 | 50.4 | 71.3 | 52 | 46.6 | 36.2 | 29.2 | 67.3 | — | — | |
| oracleBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN2020.03 | 50.4 | 49.8 | 50.6 | 40.2 | 38.9 | 53.3 | 69.4 | — | — | |
| I³NetAblation=w/o RJCA2021.03 | 50.2 | 81.8 | 46.3 | 40.4 | 33.3 | 34 | 65.1 | — | — | |
| HTCN†2021.03 | 50.1 | 78.6 | 47.5 | 45.6 | 35.4 | 31 | 62.2 | — | — | |
| MI-maxNet=RES-152-COCO2018.10 | 50.1 | 85.2 | 48.2 | 49.2 | 31 | 30 | 57 | — | — | |
| BSR+WST2021.03 | 49.9 | 75.6 | 45.8 | 49.3 | 34.1 | 30.3 | 64.1 | — | — | |
| ADDAbaseline detector=SSD3002018.03 | 49.8 | 79.9 | 49.5 | 39.5 | 35.3 | 29.4 | 65.1 | — | — | |
| Strong-Weak Domain AlignmentGlobal Alignment (G)=true, Instance-level Alignment (I)=true2018.12 | 49.8 | 66.4 | 53.7 | 43.8 | 37.9 | 31.9 | 65.3 | — | — | |
| Strong-Weak DAG=true, I=true, CTX=false, L=false2018.12 | 49.8 | 66.4 | 53.7 | 43.8 | 37.9 | 31.9 | 65.3 | — | — | |
| ADDASetting=Unsupervised domain adaptation object detection2023.09 | 49.8 | — | — | — | — | — | — | — | — | |
| Baselinebaseline detector=SSD3002018.03 | 49.6 | 79.8 | 49.5 | 38.1 | 35.1 | 30.4 | 65.1 | — | — | |
| SWDA+2021.03 | 49.5 | 73.9 | 48.6 | 44.3 | 36.2 | 31.7 | 62.1 | — | — | |
| WST2021.03 | 49.2 | 77.8 | 48 | 45.2 | 30.4 | 29.5 | 64.2 | — | — | |
| Ensemblebaseline detector=SSD3002018.03 | 48.6 | 79.8 | 49.6 | 38.1 | 35.2 | 30.4 | 58.7 | — | — | |
| BSR2021.03 | 48.6 | 82.8 | 43.2 | 49.8 | 29.6 | 27.6 | 58.4 | — | — | |
| Source Only2021.03 | 47.1 | 77.5 | 46.1 | 44.6 | 30 | 26 | 58.6 | — | — | |
| SourceBackbone=ResNet-1012021.11 | 46.9 | — | 44.5 | 53 | 24.9 | 18.8 | 56.3 | — | 84.2 | |
| SourceBackbone=ResNet-101, Source Domain=PASCAL VOC2023.09 | 46.9 | — | 44.5 | 53 | 24.9 | 18.8 | 56.3 | — | 84.2 | |
| SourceBackbone=ResNet-101, Domain Adaptation=Pascal VOC -> Watercolor2k, Resolution=Short side 600 pixels2023.09 | 46.9 | — | 44.5 | 53 | 24.9 | 18.8 | 56.3 | — | 84.2 | |
| SourceBackbone=ResNet-1012023.09 | 46.9 | — | 44.5 | 53 | 24.9 | 18.8 | 56.3 | — | 84.2 | |
| Source OnlyBackbone=ResNet-101, Pre-training=ImageNet, Architecture=Faster R-CNN2020.03 | 46.4 | 74.3 | 49 | 35.2 | 33.9 | 25.3 | 60.6 | — | — | |
| DA-FasterGlobal Alignment (G)=true, Instance-level Alignment (I)=true2018.12 | 46 | 75.2 | 40.6 | 48 | 31.5 | 20.6 | 60 | — | — | |
| DA-FasterG=true, I=true, CTX=false, L=false2018.12 | 46 | 75.2 | 40.6 | 48 | 31.5 | 20.6 | 60 | — | — | |
| DA-FasterMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 46 | 75.2 | 40.6 | 48 | 31.5 | 20.6 | 60 | — | — | |
| BDC-Faster2018.12 | 45.5 | 68.6 | 48.3 | 47.2 | 26.5 | 21.7 | 60.5 | — | — | |
| BDC-FasterG=true, I=false, CTX=false, L=false2018.12 | 45.5 | 68.6 | 48.3 | 47.2 | 26.5 | 21.7 | 60.5 | — | — | |
| BDC-FasterMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 45.5 | 68.6 | 48.3 | 47.2 | 26.5 | 21.7 | 60.5 | — | — | |
| Faster RCNN2018.12 | 44.6 | 68.8 | 46.8 | 37.2 | 32.7 | 21.3 | 60.7 | — | — | |
| Source OnlyG=false, I=false, CTX=false, L=false2018.12 | 44.6 | 68.8 | 46.8 | 37.2 | 32.7 | 21.3 | 60.7 | — | — | |
| Source-onlyMultilabel-Conditional adversary=false, Prediction based Regularization=false2020.03 | 44.6 | 68.8 | 46.8 | 37.2 | 32.7 | 21.3 | 60.7 | — | — | |
| TADP simpleModification=M_simple(P)2023.09 | 43.5 | — | — | — | — | — | — | 72.2 | — | |
| TADP TextualInversionModification=MTI(P)2023.09 | 43.2 | — | — | — | — | — | — | 72.2 | — | |
| TADP DreamBoothModification=MDB(P)2023.09 | 43.2 | — | — | — | — | — | — | 72.2 | — |