To Do

[maria-yampolskaya]

• [x] Compare results to version without secondary types
• [x] Apply pre-built neural net (transfer learning)
• [x] Use Keras Tuner
• [x] Save everything as jpgs

[Sevans711]

• [ ] Long-term goal: Multi-label network (see e.g. here)
• [x] Meet on Dec 1 at 2
• [x] Minimum for next time: make a neural network that does the basic thing we want to do
• [x] Data Augmentation (include rotated / stretched / faded copies of images in initial dataset.)