SVHN TensorFlow: Study materials, questions and answers, examples and source code related to work with The Street View House Numbers Dataset in TensorFlow.
The Street View House Numbers (SVHN) Dataset: http://ufldl.stanford.edu/housenumbers/ (Non-commercial use only) [“SVHN is a real-world image dataset for developing machine learning and object recognition algorithms with minimal requirement on data preprocessing and formatting.”]
Convolutional Neural Networks (tensorflow tutorial): https://www.tensorflow.org/versions/r0.10/tutorials/deep_cnn/index.html [‘Fork the CIFAR-10 tutorial and swap in the SVHN as the input data’]
How to get SVHN into tensorflow: https://groups.google.com/a/tensorflow.org/forum/#!topic/discuss/2TGYnPQP5hE
SVHN based on the CIFAR10 TensorFlow example: https://github.com/jkschin/svhn
SVHN TensorFlow Source code example: https://github.com/ppwwyyxx/tensorpack/blob/master/tensorpack/dataflow/dataset/svhn.py
SVHN TensorFlow Example: http://jkschin.github.io/2016/01/13/svhn-tensorflow.html
Issue with “TensorFlow Indices are not valid (out of bounds)”: http://stackoverflow.com/questions/34759227/tensorflow-cifar10-example/34763642#34763642 (‘Given 10 classes, the labels should range 0-9. The error happened because the labels ranged 1-10.’)
Reshaping SVHN data for TensorFlow: http://stackoverflow.com/questions/36384618/tensorflow-getting-data-into-it-svhn
Source code, “SVHN-classification-using-Tensorflow”: https://github.com/codemukul95/SVHN-classification-using-Tensorflow
Converting “.mat” to “.bin” with fix to labels (1-10 becomes 0-9): http://stackoverflow.com/questions/35756236/tensorflow-tried-svhn-by-editing-label-0-9-still-not-working
Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks [Paper, PDF]: http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/42241.pdf [‘recognizing arbitrary multi-digit numbers from Street View imagery‘]
Udacity Deep Learning Capstone Project: https://docs.google.com/document/d/1L11EjK0uObqjaBhHNcVPxeyIripGHSUaoEWGypuuVtk/pub [‘In this project, you will train a model that can decode sequences of digits from natural images, and create an app that prints the numbers it sees in real time. You may choose to implement your project as a simple Python script, a web app/service or an Android app.‘]
Number plate recognition with Tensorflow: http://matthewearl.github.io/2016/05/06/cnn-anpr/ (Source code: Deep ANPR)
SUN Database: Scene Categorization Benchmark: http://vision.princeton.edu/projects/2010/SUN/ [‘Scene categorization is a fundamental problem in computer vision. Scene UNderstanding (SUN) database contains 899 categories and 130,519 images.’]
TensorFlow discuss: https://groups.google.com/a/tensorflow.org/forum/#!forum/discuss [‘General discussions about TensorFlow development and directions’]
TensorFlow Tagged Questions: https://stackoverflow.com/questions/tagged/tensorflow
A curated list of deep learning resources for computer vision: https://github.com/kjw0612/awesome-deep-vision
A curated list of awesome computer vision resources: https://github.com/jbhuang0604/awesome-computer-vision
Deep Learning Capstone Project (FedorSulaev/DeepLearningCameraApp): https://github.com/FedorSulaev/DeepLearningCameraApp
Ian Goodfellow: Multi-digit Number Recognition from Street View Imagery using ConvNets:
2015 CS224d Lectures:
Please let me know if you have any other examples of SVHN TensorFlow implementation.
Subscribe here
for curated list of news, papers and sources on AI and ML