Hotdog or not hotdog
A simple hotdog/not hotdog classifier built through transfer learning on InceptionV3. In order to learn about fine-tuning a well known architecture, I decided to begin with the simplest framework, Tensorflow for Poets.
All of the training images were taken from Imagenet (Princeton University and Stanford University).
Accuracy
98-99% accuracy after being trained on 1400 images (1200 hotdog, 200 not hotdog).
How to run
In the directory, run python label_image.py hotdog.jpg (or any other image you would like to test).