Use the Inception image classification model to create embeddings of flower images. Convert these embeddings to a 2d space using TSNe and plot them with matplotlib.
- Download the flower images at: http://download.tensorflow.org/example_images/flower_photos.tgz
- Extract the bottleneck.tar.gz
- Run tsne_embedding_bottlenecks.py
If you want to create your own embeddings, follow the tutorial at: https://www.tensorflow.org/tutorials/image_retraining (don't forget to run .configure before compiling with bazel). Run the retrain script with your own images and use that image and bottleneck dir in the tsne_embedding_bottlenecks.py script
Daisy vs Tulips:
Roses vs Sunflowers:
