Domain Generalization On Vizwiz

评估指标

Accuracy - All Images

Accuracy - Clean Images

Accuracy - Corrupted Images

评测结果

各个模型在此基准测试上的表现结果

				Paper Title	Repository
VOLO-D5	57.2	59.7	51.8	VOLO: Vision Outlooker for Visual Recognition
ConvNeXt-B	53.5	56	46.9	A ConvNet for the 2020s
ResNeXt-101 32x16d	51.7	54.8	48.1	Aggregated Residual Transformations for Deep Neural Networks
EfficientNet-B8 (advprop+autoaug)	50.5	53.2	45.8	Adversarial Examples Improve Image Recognition
EfficientNet-B7 (advprop+autoaug)	49.7	52	45	Adversarial Examples Improve Image Recognition
EfficientNet-B6 (advprop+autoaug)	49.6	53.2	44.7	Adversarial Examples Improve Image Recognition
EfficientNet-B5 (advprop+autoaug)	49.1	51.7	44	Adversarial Examples Improve Image Recognition
ViT-16/L-224	49	-	-	An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
ResNet-50 (gn)	48.9	44.4	39.1	ResNet strikes back: An improved training procedure in timm
EfficientNet-B4 (advprop+autoaug)	48.1	51.4	42.5	Adversarial Examples Improve Image Recognition
ResNet-152	47.5	51.3	43.3	Deep Residual Learning for Image Recognition
ResNet-101	46.3	50.1	40.5	Deep Residual Learning for Image Recognition
EfficientNet-B6 (autoaug)	45.8	50.7	39.3	AutoAugment: Learning Augmentation Strategies From Data	-
EfficientNet-B5 (autoaug)	45.7	50.2	39.8	AutoAugment: Learning Augmentation Strategies From Data	-
EfficientNet-B3 (advprop+autoaug)	45.5	49.5	39.8	Adversarial Examples Improve Image Recognition
EfficientNet-B7 (autoaug)	45	49.9	39.1	AutoAugment: Learning Augmentation Strategies From Data	-
EfficientNet-B7 (randaug)	45	48.7	38.9	RandAugment: Practical automated data augmentation with a reduced search space
EfficientNet-B4 (autoaug)	44.3	48.6	38.2	AutoAugment: Learning Augmentation Strategies From Data	-
EfficientNet-B2 (advprop+autoaug)	44.3	48	38.2	Adversarial Examples Improve Image Recognition
ResNet-50	42.9	47.7	37.1	Deep Residual Learning for Image Recognition

0 of 90 row(s) selected.

Command Palette

Domain Generalization On Vizwiz

评估指标

评测结果