Home News Latest Papers Tutorials Datasets Wiki SOTA LLM Models GPU Leaderboard Events

English

Fine Grained Image Classification On Caltech

Metrics

Top-1 Error Rate

Results

Performance results of various models on this benchmark

Model Name	Top-1 Error Rate	Paper Title	Repository
ViT-S/16 (RPE w/ GAB)	9.798%	Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields
ResNeXt-101-32x8d	4.42%	Dead Pixel Test Using Effective Receptive Field
NNCLR	8.7%	With a Little Help from My Friends: Nearest-Neighbor Contrastive Learning of Visual Representations
SE-ResNet-101 (SAP)	15.949%	Stochastic Subsampling With Average Pooling	-
Wide-ResNet-101 (Spinal FC)	2.68%	SpinalNet: Deep Neural Network with Gradual Input
Wide-ResNet-101	2.89%	SpinalNet: Deep Neural Network with Gradual Input
Bamboo (ViT-B/16)	-	Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy
ResNet-101 (ideal number of groups)	22.247%	On the Ideal Number of Groups for Isometric Gradient Propagation	-
AutoAugment	13.07%	AutoAugment: Learning Augmentation Policies from Data
SEER (RegNet10B - linear eval)	9.0%	Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision
VIT-L/16	1.98%	Reduction of Class Activation Uncertainty with Background Information
µ2Net+ (ViT-L/16)	4.06%	A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems
PreResNet-101	15.8036%	How to Use Dropout Correctly on Residual Networks with Batch Normalization
TWIST (ResNet-50 )	6.5%	Self-Supervised Learning by Estimating Twin Class Distributions	-
Pre trained wide-resnet-101	-	ProgressiveSpinalNet architecture for FC layers
µ2Net (ViT-L/16)	7%	An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems
UL-Hopfield (ULH)	-	Unsupervised Learning using Pretrained CNN and Associative Memory Bank	-
VGG-19bn (Spinal FC)	6.84%	SpinalNet: Deep Neural Network with Gradual Input

0 of 18 row(s) selected.