HyperAI超神经

Fine Grained Image Classification On Oxford 2

评估指标

Accuracy

评测结果

各个模型在此基准测试上的表现结果

模型名称
Accuracy
Paper TitleRepository
ResNet-50-SAM91.6When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
µ2Net (ViT-L/16)95.3An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems
Assemble-ResNet-FGVC-5094.3%Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network
NAT-M394.1Neural Architecture Transfer
Mixer-B/16- SAM92.5When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
ResNet-101 (ideal number of groups)77.076On the Ideal Number of Groups for Isometric Gradient Propagation-
NAT-M293.5Neural Architecture Transfer
PreResNet-10185.5897How to Use Dropout Correctly on Residual Networks with Batch Normalization
Mixer-S/16- SAM88.7When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
SE-ResNet-101 (SAP)86.011Stochastic Subsampling With Average Pooling-
µ2Net+ (ViT-L/16)95.5A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems
EffNet-L2 (SAM)97.10Sharpness-Aware Minimization for Efficiently Improving Generalization
NAT-M494.3Neural Architecture Transfer
ViT-B/16- SAM93.1When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
BiT-L (ResNet)96.62Big Transfer (BiT): General Visual Representation Learning
ViT-S/16- SAM92.9When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
ResNet-152-SAM93.3When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations
BiT-M (ResNet)94.47Big Transfer (BiT): General Visual Representation Learning
ViT-B/16-An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
0 of 19 row(s) selected.