ML Research Wiki / Benchmarks / Fine-Grained Image Classification / Caltech-101

Caltech-101

Fine-Grained Image Classification Benchmark

Performance Over Time

📊 Showing 18 results | 📏 Metric: Top-1 Error Rate

Top Performing Models

Rank	Model	Paper	Top-1 Error Rate	Date	Code
1	Pre trained wide-resnet-101	ProgressiveSpinalNet architecture for FC layers	97.76	2021-03-21	📦 praveenchopra/ProgressiveSpinalNet
2	Wide-ResNet-101 (Spinal FC) 📚	SpinalNet: Deep Neural Network with Gradual Input	97.32	2020-07-07	📦 dipuk0506/SpinalNet 📦 dipuk0506/uq 📦 Mechachleopteryx/SpinalNet
3	ResNeXt-101-32x8d	Dead Pixel Test Using Effective Receptive Field	95.58	2021-08-31	📦 kmbmjn/deadpixeltest_code
4	Bamboo (ViT-B/16) 📚	Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy	94.80	2022-03-15	📦 zhangyuanhan-ai/bamboo 📦 davidzhangyuanhan/bamboo
5	SEER (RegNet10B - linear eval) 📚	Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision	91.00	2022-02-16	📦 facebookresearch/vissl
6	UL-Hopfield (ULH)	Unsupervised Learning using Pretrained CNN and Associative Memory Bank	91.00	2018-05-02	-
7	VIT-L/16	Reduction of Class Activation Uncertainty with Background Information	0.00	2023-05-05	📦 dipuk0506/SpinalNet 📦 dipuk0506/uq
8	Wide-ResNet-101 📚	SpinalNet: Deep Neural Network with Gradual Input	0.00	2020-07-07	📦 dipuk0506/SpinalNet 📦 dipuk0506/uq 📦 Mechachleopteryx/SpinalNet
9	µ2Net+ (ViT-L/16)	A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems	0.00	2022-09-15	📦 google-research/google-research
10	TWIST (ResNet-50 )	Self-Supervised Learning by Estimating Twin Class Distributions	0.00	2021-10-14	📦 bytedance/TWIST 📦 beresandras/contrastive-classification-keras

All Papers (18)

ProgressiveSpinalNet architecture for FC layers

2021

Pre trained wide-resnet-101

praveenchopra/ProgressiveSpinalNet

SpinalNet: Deep Neural Network with Gradual Input

2020

Wide-ResNet-101 (Spinal FC)

dipuk0506/SpinalNet dipuk0506/uq Mechachleopteryx/SpinalNet

Dead Pixel Test Using Effective Receptive Field

2021

ResNeXt-101-32x8d

kmbmjn/deadpixeltest_code

Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy

2022

Bamboo (ViT-B/16)

zhangyuanhan-ai/bamboo davidzhangyuanhan/bamboo

Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision

2022

SEER (RegNet10B - linear eval)

facebookresearch/vissl

Unsupervised Learning using Pretrained CNN and Associative Memory Bank

2018

UL-Hopfield (ULH)

Reduction of Class Activation Uncertainty with Background Information

2023

VIT-L/16

dipuk0506/SpinalNet dipuk0506/uq

SpinalNet: Deep Neural Network with Gradual Input

2020

Wide-ResNet-101

dipuk0506/SpinalNet dipuk0506/uq Mechachleopteryx/SpinalNet

A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems

2022

µ2Net+ (ViT-L/16)

google-research/google-research

Self-Supervised Learning by Estimating Twin Class Distributions

2021

TWIST (ResNet-50 )

bytedance/TWIST beresandras/contrastive-classification-keras

SpinalNet: Deep Neural Network with Gradual Input

2020

VGG-19bn (Spinal FC)

dipuk0506/SpinalNet dipuk0506/uq Mechachleopteryx/SpinalNet

An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems

2022

µ2Net (ViT-L/16)

google-research/google-research

With a Little Help from My Friends: Nearest-Neighbor Contrastive Learning of Visual Representations

2021

NNCLR

lightly-ai/lightly keras-team/keras-io

Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields

2023

ViT-S/16 (RPE w/ GAB)

kmbmjn/GaussianAttentionBias

AutoAugment: Learning Augmentation Policies from Data

2018

AutoAugment

tensorflow/models tensorflow/models

How to Use Dropout Correctly on Residual Networks with Batch Normalization

2023

PreResNet-101

kmbmjn/DropoutCorrectly

Stochastic Subsampling With Average Pooling

2024

SE-ResNet-101 (SAP)

On the Ideal Number of Groups for Isometric Gradient Propagation

2023

ResNet-101 (ideal number of groups)

Caltech-101

Performance Over Time

Edit Benchmark Results

Edit Result

Top Performing Models

All Papers (18)

ProgressiveSpinalNet architecture for FC layers

SpinalNet: Deep Neural Network with Gradual Input

Dead Pixel Test Using Effective Receptive Field

Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy

Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision

Unsupervised Learning using Pretrained CNN and Associative Memory Bank

Reduction of Class Activation Uncertainty with Background Information

SpinalNet: Deep Neural Network with Gradual Input

A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems

Self-Supervised Learning by Estimating Twin Class Distributions

SpinalNet: Deep Neural Network with Gradual Input

An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems

With a Little Help from My Friends: Nearest-Neighbor Contrastive Learning of Visual Representations

Understanding Gaussian Attention Bias of Vision Transformers Using Effective Receptive Fields

AutoAugment: Learning Augmentation Policies from Data

How to Use Dropout Correctly on Residual Networks with Batch Normalization

Stochastic Subsampling With Average Pooling

On the Ideal Number of Groups for Isometric Gradient Propagation

Model	Paper	Top-1 Error Rate	Date
Pre trained wide-resnet-101	ProgressiveSpinalNet architecture for FC layers	97.76	2021-03-21
Wide-ResNet-101 (Spinal FC)	SpinalNet: Deep Neural Network with Gradual Input	97.32	2020-07-07
ResNeXt-101-32x8d	Dead Pixel Test Using Effective Receptive Field	95.58	2021-08-31
Bamboo (ViT-B/16)	Bamboo: Building Mega-Scale Vision Dataset Contin…	94.80	2022-03-15
SEER (RegNet10B - linear eval)	Vision Models Are More Robust And Fair When Pretr…	91.00	2022-02-16
UL-Hopfield (ULH)	Unsupervised Learning using Pretrained CNN and As…	91.00	2018-05-02
VIT-L/16	Reduction of Class Activation Uncertainty with Ba…		2023-05-05
Wide-ResNet-101	SpinalNet: Deep Neural Network with Gradual Input		2020-07-07
µ2Net+ (ViT-L/16)	A Continual Development Methodology for Large-sca…		2022-09-15
TWIST (ResNet-50 )	Self-Supervised Learning by Estimating Twin Class…		2021-10-14
VGG-19bn (Spinal FC)	SpinalNet: Deep Neural Network with Gradual Input		2020-07-07
µ2Net (ViT-L/16)	An Evolutionary Approach to Dynamic Introduction …		2022-05-25
NNCLR	With a Little Help from My Friends: Nearest-Neigh…		2021-04-29
ViT-S/16 (RPE w/ GAB)	Understanding Gaussian Attention Bias of Vision T…		2023-05-08
AutoAugment	AutoAugment: Learning Augmentation Policies from …		2018-05-24
PreResNet-101	How to Use Dropout Correctly on Residual Networks…		2023-02-13
SE-ResNet-101 (SAP)	Stochastic Subsampling With Average Pooling		2024-09-25
ResNet-101 (ideal number of groups)	On the Ideal Number of Groups for Isometric Gradi…		2023-02-07