Image Classification On Inaturalist 2019

评估指标

Top-1 Accuracy

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
Hiera-H (448px)	88.5	Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles
MAE (ViT-H, 448)	88.3	Masked Autoencoders Are Scalable Vision Learners
Grafit (RegnetY 8GF)	84.1	Grafit: Learning fine-grained image representations with coarse labels	-
MixMIM-L	83.9	MixMAE: Mixed and Masked Autoencoder for Efficient Pretraining of Hierarchical Vision Transformers
RDNet-L (224 res, IN-1K pretrained)	83.7	DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
RDNet-B (224 res, IN-1K pretrained)	83.5	DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
RDNet-S (224 res, IN-1K pretrained)	82.9	DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
Conviformer-B	82.85	Conviformers: Convolutionally guided Vision Transformer
CeiT-S (384 finetune resolution)	82.7	Incorporating Convolution Designs into Visual Transformers
CaiT-M-36 U 224	81.8	-	-
RDNet-T (224 res, IN-1K pretrained)	81.2	DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
CeiT-S	78.9	Incorporating Convolution Designs into Visual Transformers
CeiT-T (384 finetune resolution)	77.9	Incorporating Convolution Designs into Visual Transformers
ResNet50 (A2)	75.0	ResNet strikes back: An improved training procedure in timm
LeViT-384	74.3	LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference
CeiT-T	72.8	Incorporating Convolution Designs into Visual Transformers
ResMLP-24	72.5	ResMLP: Feedforward networks for image classification with data-efficient training
LeViT-256	72.3	LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference
ResMLP-12	71.0	ResMLP: Feedforward networks for image classification with data-efficient training
LeViT-192	70.8	LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference

0 of 22 row(s) selected.