Object Detection on MS-COCO 5% labeled protocol
32.1mAPDual Pseudo-label Polishing Framework
Evaluation Results
| Method | Links | |
|---|---|---|
| Dual Pseudo-label Polishing Framework2022.07 | 32.1 | |
| Virtual Category Learning (with tricks)Configuration=Training tricks in Soft Teacher2022.07 | 32.05 | |
| Soft Teacher2022.07 | 30.74 | |
| Soft TeacherRemark=ICCV 20212022.07 | 30.74 | |
| Virtual Category LearningConfiguration=Same settings as Unbiased Teacher2022.07 | 30.31 | |
| Region Uncertainty QuantificationStrong data augmentation=true2021.11 | 28.96 | |
| Combating Noise2022.07 | 28.96 | |
| Unbiased Teacher2022.07 | 28.27 | |
| Unbiased TeacherRemark=ICLR 20212022.07 | 28.27 | |
| Humble TeacherStrong data augmentation=true2021.11 | 27.7 | |
| Humble Teacher2022.07 | 27.7 | |
| Humble TeacherRemark=CVPR 20212022.07 | 27.7 | |
| Instant-Teaching*Backbone=R50-FPN, co-rectify scheme=true2021.03 | 26.75 | |
| Instant-TeachingStrong data augmentation=true2021.11 | 26.75 | |
| Instant Teaching2022.07 | 26.75 | |
| Instant-Teaching*Remark=CVPR 20212022.07 | 26.75 | |
| Interactive2022.07 | 26.37 | |
| Unbiased TeacherStrong data augmentation=true2021.11 | 26.3 | |
| Instant-TeachingBackbone=R50-FPN, co-rectify scheme=false2021.03 | 25.5 | |
| Instant-TeachingRemark=CVPR 20212022.07 | 25.5 | |
| STACBackbone=ResNet-50 with FPN, Augmentation=C+{B,G}+Cutout, Training Data=Semi-supervised2020.05 | 24.38 | |
| STACBackbone=R50-FPN2021.03 | 24.38 | |
| STACStrong data augmentation=true2021.11 | 24.38 | |
| STAC2022.07 | 24.38 | |
| STACRemark=arxiv 20202022.07 | 24.38 | |
| Supervised†Backbone=ResNet-50 with FPN, Augmentation=C+{B,G}+Cutout, Training Data=Labeled data only2020.05 | 21.18 | |
| Supervised2022.07 | 20.92 | |
| CSD†Backbone=R50-FPN2021.03 | 18.9 | |
| CSDStrong data augmentation=false2021.11 | 18.63 | |
| CSD2022.07 | 18.63 | |
| SupervisedBackbone=ResNet-50 with FPN, Augmentation=Standard, Training Data=Labeled data only2020.05 | 18.47 | |
| SupervisedBackbone=R50-FPN2021.03 | 18.47 | |
| Supervised2022.07 | 18.47 |