SVHN
A real-world image dataset for machine learning and object recognition, derived from Google Street View house numbers, with over 600,000 labeled digit images.
Score
30
User Rating
3.0
Free/Trial Support
Not Supported
Starting Price
--
Last Updated
Apr 23, 2026
Overview
What is SVHN?
SVHN is a dataset of house number images from Google Street View, designed for developing machine learning and object recognition algorithms. It includes over 600,000 labeled digits in two formats: original images with bounding boxes and MNIST-like 32x32 cropped images.
What is SVHN used for?
Top Features
- Over 600,000 labeled digit images
- Two formats: original with bounding boxes and 32x32 cropped
- Real-world data from Google Street View
- Minimal preprocessing required
- Includes training, testing, and extra data
How to use SVHN?
Download the dataset in either original or cropped format, load the files (e.g., .mat files for cropped images), and use the images and labels for training and testing machine learning models for digit recognition tasks.
Alternative Tools
No alternative tools found
Pros & Cons
Pros
- Large dataset size
- Real-world applicability
- Multiple data formats
- Free for non-commercial use
- Well-documented with references
Cons
- Non-commercial use only for downloads
- Some distracting elements in cropped images
- Requires specific tools like Matlab for some formats
Use Cases
- Training machine learning models for digit recognition
- Academic research in computer vision
- Benchmarking object recognition algorithms
- Educational purposes in AI courses
User Reviews
3.0
out of 5
3.0
No reviews yet
Be the first to review this tool!