Training, data preparation, scraping scripts:
- I created https://github.com/rolba/rScraper API for downloading data from the internet for training purposes. Created for this project.
- aimaschroomerprepare.py - File that generates sugumented data for training purposes.
- aimaschroomersplitter.py - Splits data into train set, validate set and test set.
- aimaschroomerhdf5generator.py - Script that generates HDF5 file containers for thraining purposes.
- aimaschroomermodelserializer.py - Converts saved weights to serialized mode.
- aimaschroomertrain.py - Trains MobileNet deep neural network and uses data generated in HDF5 containers.
Helpers:
- trainingmonitor.py - Helps to take care of training process - I can easily find out that I overfit by printing plots per every epoch.
- DataGenerator.py - helper class for preprocessing images before they are send to hdf5 data file containers. (Reused)
- DatasetWriter.py - helper class that puts images into hdf5 data file container. (Reused)
- mobilenet.py - Taken from Keras Repo.
Main App.
- main.py - Main GUI application file. Runs propoer classes.
- mobilenet.py - Logic class used for creating GUI, and all logic that is needed for taking a photo, passing it to model and make classification if muschroom is Eatable or poisoned.