SVHN (Street View House Numbers)

Introduced by Netzer et al. in Reading digits in natural images with unsupervised feature learning

Street View House Numbers (SVHN) is a digit classification benchmark dataset that contains 600,000 32×32 RGB images of printed digits (from 0 to 9) cropped from pictures of house number plates. The cropped images are centered in the digit of interest, but nearby digits and other distractors are kept in the image. SVHN has three sets: training, testing sets and an extra set with 530,000 images that are less difficult and can be used for helping with the training process.

Source: Competitive Multi-scale Convolution

Homepage

Benchmarks

Add a new result Link an existing benchmark

Task	Dataset Variant	Best Model
Image Classification	SVHN	WRN28-10
Semi-Supervised Image Classification	SVHN, 1000 labels	Meta Pseudo Labels
Semi-Supervised Image Classification	SVHN, 250 Labels	ShrinkMatch
Domain Adaptation	SVHN-to-MNIST	Mean teacher
Semi-Supervised Image Classification	SVHN, 500 Labels	MixMatch
Semi-Supervised Image Classification	SVHN, 40 Labels	ShrinkMatch
Unsupervised Image Classification	SVHN	ACOL-GAR
Semi-Supervised Image Classification	SVHN, 2000 Labels	MixMatch
Semi-Supervised Image Classification	SVHN, 4000 Labels	MixMatch
Sparse Representation-based Classification	SVHN	DSRC
Novel Class Discovery	SVHN	AutoNovel
Anomaly Detection	SVHN	RCALAD