yinleon / pydata2017

This is the code and presentation for my PyData2017 talk "Reverse Image Search Using Out-of-the-box Machine Learning Libraries

Reverse Image Search

This repo is for my presentation at PyData 2017.

It demonstrates how to utilize Keras' pretrained ResNet50 convolutional neural network and scikit-learns K-nearest neighbors for a reverse image search enginer.

There is no requirements.txt file, becuase configuring tensorflow and keras can be finicky.
I ran the code use Keras with Tensorflow and just one nvidia k80 GPU.

Here is a video.

Copyright

Copyright (c) 2018 Leon Yin. All Rights Reserved.

Research Outputs

If you use this method in your research please cite this as:

@booklet{key,
author = {Leon Yin},
title = {Reverse Image Search Using Out-of-the-box Machine Learning Libraries},
howpublished= {Paper presented at PyData 2017, New York NY},
year = {2017}
}

Conversely if this software was helpful you can cite it as:

@software{ReverseImageSearch,
  author = {Leon Yin},
  title = {ReverseImageSearch},
  year = {2017}
  howpublished = {https://github.com/yinleon/pydata2017},
}

Next Steps

I am trying to build a UI for this to be used at Data & Society and journalists.

Conclusion

Thanks for stopping by! Please contact me over social media or elsewhere if you have any questions, concerns, or post an Issue if anything looks funky.