List your company
ResearchFormerly WizWand

Object Detection on COCO 2017 (train val)

40.1AP (IoU 0.5:0.95)SMoG
4.032813.396422.7632.1236Jul 11, 2021Sep 10, 2021Nov 10, 2021Jan 10, 2022Mar 12, 2022May 12, 2022Jul 13, 2022
Updated 8mo ago

Evaluation results

MethodLinks
SMoGModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0740.161.643.7
MoCoV2Model=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0739.459.943
BYOLModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0739.459.943
Barlow TwinsModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0739.258.742.6
SupervisedModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0738.959.642.7
PIRLModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0738.557.641.2
SwAVModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0738.560.441.4
DC-v2Model=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0738.360.341.3
InsDisModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.0737.457.640.6
Adaptive Class-Rebalancing Self-TrainingLabeled data percentage=10%2021.0734.92——
Unbiased Teacher v2Labeled data percentage=10%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0632.61——
Unbiased TeacherLabeled data percentage=10%2021.0731.5——
Adaptive Class-Rebalancing Self-TrainingLabeled data percentage=5%2021.0731.35——
Rand InitModel=Mask R-CNN, Backbone=ResNet-50-FPN, Evaluation Protocol=fine-tuning2022.073149.533.2
Instant TeachingLabeled data percentage=10%2021.0730.4——
Unbiased Teacher v2Labeled data percentage=5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0630.08——
Adaptive Class-Rebalancing Self-TrainingLabeled data percentage=2%2021.0728.69——
STACLabeled data percentage=10%2021.0728.64——
Unbiased TeacherLabeled data percentage=5%2021.0728.27——
Unbiased TeacherLabeled data percentage=10%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0628.18——
Instant TeachingLabeled data percentage=5%2021.0726.75——
Adaptive Class-Rebalancing Self-TrainingLabeled data percentage=1%2021.0726.07——
Unbiased Teacher v2Labeled data percentage=2%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0626.03——
STACLabeled data percentage=5%2021.0724.38——
STACLabeled data percentage=10%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0624.31——
Unbiased TeacherLabeled data percentage=2%2021.0724.3——
Unbiased TeacherLabeled data percentage=5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0623.99——
SupervisedLabeled data percentage=10%2021.0723.86——
Unbiased Teacher v2Labeled data percentage=1%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0622.71——
CSDLabeled data percentage=10%2021.0722.46——
Instant TeachingLabeled data percentage=2%2021.0722.45——
CSDLabeled data percentage=10%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0622.06——
SupervisedLabeled data percentage=10%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0620.98——
Unbiased TeacherLabeled data percentage=1%2021.0720.75——
STACLabeled data percentage=5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0620.36——
Adaptive Class-Rebalancing Self-TrainingLabeled data percentage=0.5%2021.0719.62——
Unbiased TeacherLabeled data percentage=2%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0618.7——
CSDLabeled data percentage=5%2021.0718.63——
SupervisedLabeled data percentage=5%2021.0718.47——
STACLabeled data percentage=2%2021.0718.25——
CSDLabeled data percentage=5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0618.09——
Instant TeachingLabeled data percentage=1%2021.0718.05——
SupervisedLabeled data percentage=5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0617.01——
Unbiased TeacherLabeled data percentage=0.5%2021.0716.94——
Unbiased Teacher v2Labeled data percentage=0.5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0616.25——
STACLabeled data percentage=2%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0615.5——
Unbiased TeacherLabeled data percentage=1%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0614.61——
STACLabeled data percentage=1%2021.0713.97——
CSDLabeled data percentage=2%2021.0713.93——
SupervisedLabeled data percentage=2%2021.0712.7——
CSDLabeled data percentage=2%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0612.53——
SupervisedLabeled data percentage=2%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.0611.97——
STACLabeled data percentage=1%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0611.97——
CSDLabeled data percentage=1%2021.0710.51——
Unbiased TeacherLabeled data percentage=0.5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.0610.27——
STACLabeled data percentage=0.5%2021.079.78——
CSDLabeled data percentage=1%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.069.23——
SupervisedLabeled data percentage=1%2021.079.05——
STACLabeled data percentage=0.5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.068.79——
SupervisedLabeled data percentage=1%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.068.43——
CSDLabeled data percentage=0.5%2021.077.41——
SupervisedLabeled data percentage=0.5%2021.076.83——
CSDLabeled data percentage=0.5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/8, Reimplemented=true2022.065.76——
SupervisedLabeled data percentage=0.5%, Backbone=ResNet-50, Detector=FCOS, Batch size (labeled/unlabeled)=8/82022.065.42——