数据集

• About (opens new window)
• Datasets (opens new window)
• Contact (opens new window)

Classification datasets results

---

What is the class of this image ?

Discover the current state of the art in objects classification.

• MNIST (opens new window)
• CIFAR-10 (opens new window)
• CIFAR-100 (opens new window)
• STL-10 (opens new window)
• SVHN (opens new window)
• ILSVRC2012 task 1 (opens new window)

MNIST who is the best in MNIST ?

• (opens new window)

MNIST (opens new window) 50 results collected

Units: error %

Classify handwriten digits (opens new window). Some additional results are available on the original dataset page (opens new window).

Result	Method	Venue	Details
0.21%	Regularization of Neural Networks using DropConnect (opens new window)	ICML 2013
0.23%	Multi-column Deep Neural Networks for Image Classification (opens new window)	CVPR 2012
0.23%	APAC: Augmented PAttern Classification with Neural Networks (opens new window)	arXiv 2015
0.24%	Batch-normalized Maxout Network in Network (opens new window)	arXiv 2015	Details
0.29%	Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree (opens new window)	AISTATS 2016	Details
0.31%	Recurrent Convolutional Neural Network for Object Recognition (opens new window)	CVPR 2015
0.31%	On the Importance of Normalisation Layers in Deep Learning with Piecewise Linear Activation Units (opens new window)	arXiv 2015
0.32%	Fractional Max-Pooling (opens new window)	arXiv 2015	Details
0.33%	Competitive Multi-scale Convolution (opens new window)	arXiv 2015
0.35%	Deep Big Simple Neural Nets Excel on Handwritten Digit Recognition (opens new window)	Neural Computation 2010	Details
0.35%	C-SVDDNet: An Effective Single-Layer Network for Unsupervised Feature Learning (opens new window)	arXiv 2014
0.37%	Enhanced Image Classification With a Fast-Learning Shallow Convolutional Neural Network (opens new window)	arXiv 2015	Details
0.39%	Efficient Learning of Sparse Representations with an Energy-Based Model (opens new window)	NIPS 2006	Details
0.39%	Convolutional Kernel Networks (opens new window)	arXiv 2014	Details
0.39%	Deeply-Supervised Nets (opens new window)	arXiv 2014
0.4%	Best Practices for Convolutional Neural Networks Applied to Visual Document Analysis (opens new window)	Document Analysis and Recognition 2003
0.40%	Hybrid Orthogonal Projection and Estimation (HOPE): A New Framework to Probe and Learn Neural Networks (opens new window)	arXiv 2015
0.42%	Multi-Loss Regularized Deep Neural Network (opens new window)	CSVT 2015	Details
0.45%	Maxout Networks (opens new window)	ICML 2013	Details
0.45%	Training Very Deep Networks (opens new window)	NIPS 2015	Details
0.45%	ReNet: A Recurrent Neural Network Based Alternative to Convolutional Networks (opens new window)	arXiv 2015
0.46%	Deep Convolutional Neural Networks as Generic Feature Extractors (opens new window)	IJCNN 2015	Details
0.47%	Network in Network (opens new window)	ICLR 2014	Details
0.52 %	Trainable COSFIRE filters for keypoint detection and pattern recognition (opens new window)	PAMI 2013	Details
0.53%	What is the Best Multi-Stage Architecture for Object Recognition? (opens new window)	ICCV 2009	Details
0.54%	Deformation Models for Image Recognition (opens new window)	PAMI 2007	Details
0.54%	A trainable feature extractor for handwritten digit recognition (opens new window)	Journal Pattern Recognition 2007	Details
0.56%	Training Invariant Support Vector Machines (opens new window)	Machine Learning 2002	Details
0.59%	Simple Methods for High-Performance Digit Recognition Based on Sparse Coding (opens new window)	TNN 2008	Details
0.62%	Unsupervised learning of invariant feature hierarchies with applications to object recognition (opens new window)	CVPR 2007	Details
0.62%	PCANet: A Simple Deep Learning Baseline for Image Classification? (opens new window)	arXiv 2014	Details
0.63%	Shape matching and object recognition using shape contexts (opens new window)	PAMI 2002	Details
0.64%	Beyond Spatial Pyramids: Receptive Field Learning for Pooled Image Features (opens new window)	CVPR 2012
0.68%	Handwritten Digit Recognition using Convolutional Neural Networks and Gabor Filters (opens new window)	ICCI 2003
0.69%	On Optimization Methods for Deep Learning (opens new window)	ICML 2011
0.71%	Deep Fried Convnets (opens new window)	ICCV 2015	Details
0.75%	Sparse Activity and Sparse Connectivity in Supervised Learning (opens new window)	JMLR 2013
0.78%	Explaining and Harnessing Adversarial Examples (opens new window)	ICLR 2015	Details
0.82%	Convolutional Deep Belief Networks for Scalable Unsupervised Learning of Hierarchical Representations	ICML 2009
0.84%	Supervised Translation-Invariant Sparse Coding (opens new window)	CVPR 2010	Details
0.94%	Large-Margin kNN Classification using a Deep Encoder Network	2009
0.95%	Deep Boltzmann Machines (opens new window)	AISTATS 2009
1.01%	BinaryConnect: Training Deep Neural Networks with binary weights during propagations (opens new window)	NIPS 2015	Details
1.1%	StrongNet: mostly unsupervised image recognition with strong neurons (opens new window)	technical report on ALGLIB website 2014	Details
1.12%	CS81: Learning words with Deep Belief Networks	2008
1.19%	Convolutional Neural Networks	2003	Details
1.2%	Reducing the dimensionality of data with neural networks	2006
1.40%	Convolutional Clustering for Unsupervised Learning (opens new window)	arXiv 2015	Details
1.5%	Deep learning via semi-supervised embedding	2008
14.53%	Deep Representation Learning with Target Coding (opens new window)	AAAI 2015

Something is off, something is missing ? Feel free to fill in the form (opens new window).

CIFAR-10 who is the best in CIFAR-10 ?

• (opens new window)

CIFAR-10 (opens new window) 49 results collected

Units: accuracy %

Classify 32x32 colour images (opens new window).

Result	Method	Venue	Details
96.53%	Fractional Max-Pooling (opens new window)	arXiv 2015	Details
95.59%	Striving for Simplicity: The All Convolutional Net (opens new window)	ICLR 2015	Details
94.16%	All you need is a good init (opens new window)	ICLR 2016	Details
94%	Lessons learned from manually classifying CIFAR-10 (opens new window)	unpublished 2011	Details
93.95%	Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree (opens new window)	AISTATS 2016	Details
93.72%	Spatially-sparse convolutional neural networks (opens new window)	arXiv 2014
93.63%	Scalable Bayesian Optimization Using Deep Neural Networks (opens new window)	ICML 2015
93.57%	Deep Residual Learning for Image Recognition (opens new window)	arXiv 2015	Details
93.45%	Fast and Accurate Deep Network Learning by Exponential Linear Units (opens new window)	arXiv 2015	Details
93.34%	Universum Prescription: Regularization using Unlabeled Data (opens new window)	arXiv 2015
93.25%	Batch-normalized Maxout Network in Network (opens new window)	arXiv 2015	Details
93.13%	Competitive Multi-scale Convolution (opens new window)	arXiv 2015
92.91%	Recurrent Convolutional Neural Network for Object Recognition (opens new window)	CVPR 2015	Details
92.49%	Learning Activation Functions to Improve Deep Neural Networks (opens new window)	ICLR 2015	Details
92.45%	cifar.torch (opens new window)	unpublished 2015	Details
92.40%	Training Very Deep Networks (opens new window)	NIPS 2015	Details
92.23%	Stacked What-Where Auto-encoders (opens new window)	arXiv 2015
91.88%	Multi-Loss Regularized Deep Neural Network (opens new window)	CSVT 2015	Details
91.78%	Deeply-Supervised Nets (opens new window)	arXiv 2014	Details
91.73%	BinaryConnect: Training Deep Neural Networks with binary weights during propagations (opens new window)	NIPS 2015	Details
91.48%	On the Importance of Normalisation Layers in Deep Learning with Piecewise Linear Activation Units (opens new window)	arXiv 2015
91.40%	Spectral Representations for Convolutional Neural Networks (opens new window)	NIPS 2015
91.2%	Network In Network (opens new window)	ICLR 2014	Details
91.19%	Speeding up Automatic Hyperparameter Optimization of Deep Neural Networks by Extrapolation of Learning Curves (opens new window)	IJCAI 2015	Details
90.78%	Deep Networks with Internal Selective Attention through Feedback Connections (opens new window)	NIPS 2014	Details
90.68%	Regularization of Neural Networks using DropConnect (opens new window)	ICML 2013
90.65%	Maxout Networks (opens new window)	ICML 2013	Details
90.61%	Improving Deep Neural Networks with Probabilistic Maxout Units (opens new window)	ICLR 2014	Details
90.5%	Practical Bayesian Optimization of Machine Learning Algorithms (opens new window)	NIPS 2012	Details
89.67%	APAC: Augmented PAttern Classification with Neural Networks (opens new window)	arXiv 2015
89.14%	Deep Convolutional Neural Networks as Generic Feature Extractors (opens new window)	IJCNN 2015	Details
89%	ImageNet Classification with Deep Convolutional Neural Networks (opens new window)	NIPS 2012	Details
88.80%	Empirical Evaluation of Rectified Activations in Convolution Network (opens new window)	ICML workshop 2015	Details
88.79%	Multi-Column Deep Neural Networks for Image Classification (opens new window)	CVPR 2012	Details
87.65%	ReNet: A Recurrent Neural Network Based Alternative to Convolutional Networks (opens new window)	arXiv 2015
86.70 %	An Analysis of Unsupervised Pre-training in Light of Recent Advances (opens new window)	ICLR 2015	Details
84.87%	Stochastic Pooling for Regularization of Deep Convolutional Neural Networks (opens new window)	arXiv 2013
84.4%	Improving neural networks by preventing co-adaptation of feature detectors (opens new window)	arXiv 2012	Details
83.96%	Discriminative Learning of Sum-Product Networks (opens new window)	NIPS 2012
82.9%	Stable and Efficient Representation Learning with Nonnegativity Constraints (opens new window)	ICML 2014	Details
82.2%	Learning Invariant Representations with Local Transformations (opens new window)	ICML 2012	Details
82.18%	Convolutional Kernel Networks (opens new window)	arXiv 2014	Details
82%	Discriminative Unsupervised Feature Learning with Convolutional Neural Networks (opens new window)	NIPS 2014	Details
80.02%	Learning Smooth Pooling Regions for Visual Recognition (opens new window)	BMVC 2013
80%	Object Recognition with Hierarchical Kernel Descriptors (opens new window)	CVPR 2011
79.7%	Learning with Recursive Perceptual Representations (opens new window)	NIPS 2012	Details
79.6 %	An Analysis of Single-Layer Networks in Unsupervised Feature Learning (opens new window)	AISTATS 2011	Details
78.67%	PCANet: A Simple Deep Learning Baseline for Image Classification? (opens new window)	arXiv 2014	Details
75.86%	Enhanced Image Classification With a Fast-Learning Shallow Convolutional Neural Network (opens new window)	arXiv 2015	Details

Something is off, something is missing ? Feel free to fill in the form (opens new window).

CIFAR-100 who is the best in CIFAR-100 ?

• (opens new window)

CIFAR-100 (opens new window) 31 results collected

Units: accuracy %

Classify 32x32 colour images (opens new window).

Result	Method	Venue	Details
75.72%	Fast and Accurate Deep Network Learning by Exponential Linear Units (opens new window)	arXiv 2015	Details
75.7%	Spatially-sparse convolutional neural networks (opens new window)	arXiv 2014
73.61%	Fractional Max-Pooling (opens new window)	arXiv 2015	Details
72.60%	Scalable Bayesian Optimization Using Deep Neural Networks (opens new window)	ICML 2015
72.44%	Competitive Multi-scale Convolution (opens new window)	arXiv 2015
72.34%	All you need is a good init (opens new window)	ICLR 2015	Details
71.14%	Batch-normalized Maxout Network in Network (opens new window)	arXiv 2015	Details
70.80%	On the Importance of Normalisation Layers in Deep Learning with Piecewise Linear Activation Units (opens new window)	arXiv 2015
69.17%	Learning Activation Functions to Improve Deep Neural Networks (opens new window)	ICLR 2015	Details
69.12%	Stacked What-Where Auto-encoders (opens new window)	arXiv 2015
68.53%	Multi-Loss Regularized Deep Neural Network (opens new window)	CSVT 2015	Details
68.40%	Spectral Representations for Convolutional Neural Networks (opens new window)	NIPS 2015
68.25%	Recurrent Convolutional Neural Network for Object Recognition (opens new window)	CVPR 2015
67.76%	Training Very Deep Networks (opens new window)	NIPS 2015	Details
67.68%	Deep Convolutional Neural Networks as Generic Feature Extractors (opens new window)	IJCNN 2015	Details
67.63%	Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree (opens new window)	AISTATS 2016	Details
67.38%	HD-CNN: Hierarchical Deep Convolutional Neural Network for Large Scale Visual Recognition (opens new window)	ICCV 2015
67.16%	Universum Prescription: Regularization using Unlabeled Data (opens new window)	arXiv 2015
66.29%	Striving for Simplicity: The All Convolutional Net (opens new window)	ICLR 2014
66.22%	Deep Networks with Internal Selective Attention through Feedback Connections (opens new window)	NIPS 2014
65.43%	Deeply-Supervised Nets (opens new window)	arXiv 2014	Details
64.77%	Deep Representation Learning with Target Coding (opens new window)	AAAI 2015
64.32%	Network in Network (opens new window)	ICLR 2014	Details
63.15%	Discriminative Transfer Learning with Tree-based Priors (opens new window)	NIPS 2013	Details
61.86%	Improving Deep Neural Networks with Probabilistic Maxout Units (opens new window)	ICLR 2014
61.43%	Maxout Networks (opens new window)	ICML 2013	Details
60.8%	Stable and Efficient Representation Learning with Nonnegativity Constraints (opens new window)	ICML 2014	Details
59.75%	Empirical Evaluation of Rectified Activations in Convolution Network (opens new window)	ICML workshop 2015	Details
57.49%	Stochastic Pooling for Regularization of Deep Convolutional Neural Networks (opens new window)	arXiv 2013
56.29%	Learning Smooth Pooling Regions for Visual Recognition (opens new window)	BMVC 2013	Details
54.23%	Beyond Spatial Pyramids: Receptive Field Learning for Pooled Image Features (opens new window)	CVPR 2012

Something is off, something is missing ? Feel free to fill in the form (opens new window).

STL-10 who is the best in STL-10 ?

• (opens new window)

STL-10 (opens new window) 18 results collected

Units: accuracy %

Similar to CIFAR-10 but with 96x96 images. Original dataset website (opens new window).

Result	Method	Venue	Details
74.33%	Stacked What-Where Auto-encoders (opens new window)	arXiv 2015
74.10%	Convolutional Clustering for Unsupervised Learning (opens new window)	arXiv 2015	Details
73.15%	Deep Representation Learning with Target Coding (opens new window)	AAAI 2015
72.8% (±0.4%)	Discriminative Unsupervised Feature Learning with Convolutional Neural Networks (opens new window)	NIPS 2014	Details
70.20 % (±0.7 %)	An Analysis of Unsupervised Pre-training in Light of Recent Advances (opens new window)	ICLR 2015	Details
70.1% (±0.6%)	Multi-Task Bayesian Optimization (opens new window)	NIPS 2013	Details
68.23% ± 0.5	C-SVDDNet: An Effective Single-Layer Network for Unsupervised Feature Learning (opens new window)	arXiv 2014
68% (±0.55%)	Committees of deep feedforward networks trained with few data (opens new window)	arXiv 2014
67.9% (±0.6%)	Stable and Efficient Representation Learning with Nonnegativity Constraints (opens new window)	ICML 2014	Details
64.5% (±1%)	Unsupervised Feature Learning for RGB-D Based Object Recognition (opens new window)	ISER 2012	Details
62.32%	Convolutional Kernel Networks (opens new window)	arXiv 2014	Details
62.3% (±1%)	Discriminative Learning of Sum-Product Networks (opens new window)	NIPS 2012
61.0% (±0.58%)	No more meta-parameter tuning in unsupervised sparse feature learning (opens new window)	arXiv 2014
61%	Deep Learning of Invariant Features via Simulated Fixations in Video (opens new window)	NIPS 2012 2012
60.1% (±1%)	Selecting Receptive Fields in Deep Networks (opens new window)	NIPS 2011
58.7%	Learning Invariant Representations with Local Transformations (opens new window)	ICML 2012
58.28%	Pooling-Invariant Image Feature Learning (opens new window)	arXiv 2012	Details
56.5%	Deep Learning of Invariant Features via Simulated Fixations in Video (opens new window)	NIPS 2012	Details

Something is off, something is missing ? Feel free to fill in the form (opens new window).

SVHN who is the best in SVHN ?

• (opens new window)

SVHN (opens new window) 17 results collected

Units: error %

The Street View House Numbers (SVHN) Dataset (opens new window).

SVHN is a real-world image dataset for developing machine learning and object recognition algorithms with minimal requirement on data preprocessing and formatting. It can be seen as similar in flavor to MNIST(e.g., the images are of small cropped digits), but incorporates an order of magnitude more labeled data (over 600,000 digit images) and comes from a significantly harder, unsolved, real world problem (recognizing digits and numbers in natural scene images). SVHN is obtained from house numbers in Google Street View images.

Result	Method	Venue	Details
1.69%	Generalizing Pooling Functions in Convolutional Neural Networks: Mixed, Gated, and Tree (opens new window)	AISTATS 2016	Details
1.76%	Competitive Multi-scale Convolution (opens new window)	arXiv 2015
1.77%	Recurrent Convolutional Neural Network for Object Recognition (opens new window)	CVPR 2015	Details
1.81%	Batch-normalized Maxout Network in Network (opens new window)	arXiv 2015	Details
1.92%	Deeply-Supervised Nets (opens new window)	arXiv 2014
1.92%	Multi-Loss Regularized Deep Neural Network (opens new window)	CSVT 2015	Details
1.94%	Regularization of Neural Networks using DropConnect (opens new window)	ICML 2013
1.97%	On the Importance of Normalisation Layers in Deep Learning with Piecewise Linear Activation Units (opens new window)	arXiv 2015
2%	Estimated human performance (opens new window)	NIPS 2011	Details
2.15%	BinaryConnect: Training Deep Neural Networks with binary weights during propagations (opens new window)	NIPS 2015
2.16%	Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks (opens new window)	ICLR 2014	Details
2.35%	Network in Network (opens new window)	ICLR 2014	Details
2.38%	ReNet: A Recurrent Neural Network Based Alternative to Convolutional Networks (opens new window)	arXiv 2015
2.47%	Maxout Networks (opens new window)	ICML 2013	Details
2.8%	Stochastic Pooling for Regularization of Deep Convolutional Neural Networks (opens new window)	arXiv 2013	Details
3.96%	Enhanced Image Classification With a Fast-Learning Shallow Convolutional Neural Network (opens new window)	arXiv 2015	Details
4.9%	Convolutional neural networks applied to house numbers digit classification (opens new window)	ICPR 2012	Details

Something is off, something is missing ? Feel free to fill in the form (opens new window).

ILSVRC2012 task 1 who is the best in ILSVRC2012 task 1 ?

• (opens new window)

ILSVRC2012 task 1 (opens new window)

Units: Error (5 guesses)

1000 categories classification challenge (opens new window). With tens of thousands of training, validation and testing images.

See this interesting comparative analysis (opens new window).

Results are collected in the following external webpage (opens new window)

Last updated on 2016-02-22.

© 2013-2016 Rodrigo Benenson.

Built using middleman (opens new window) and bootstrap (opens new window).