Main script: a_or_b.py
The original project was about distinguishing horses and humans. This one is generalized solution for binary neural network which distinguishes between any given two types of objects
What it does:
- Opens pop-up window to choose folder with training data (and optionally validation data) The folder must contain two sub-folders named by the data type. (e.g humans, horses)
- Load training images from folder (shows some random examples of images)
- Preprocess images using convolutions and pooling
- Feeds training images to train a neural network
- After neural network is trained - continuously asked user to upload an image to recognize - until the user clicks "Cancel"
- Gives the result of what is on picture
Example of resources:
Training pictures for horses:
https://storage.googleapis.com/laurencemoroney-blog.appspot.com/horse-or-human.zip
Validation pictures for horses:
https://storage.googleapis.com/laurencemoroney-blog.appspot.com/horse-or-human.zip
The archive has to be unpacked and chosen in the script