TODOS:
- format the tensor to make the training net run
- split the data into training and testing groups
- break the data into batches
- improve the one-hot encoding labels
- add error handling and dynamic reading on the data folder crawler
- improve the features
- and more...