We are building a CNN to map several habitats from aerial imagery, we'll be generating samples from multiple images and storing all of the samples in project folders in a single sample folder. so the sample folder will have folders named for each project eg Image_1, Image_2, Image_3 etc We'll then have an overarching  project that will pull all of the samples together to train a model that uses all of the samples from the the individual images. Overtime we'll be adding more samples from other images to the sample folder and will train and enhance the model. Ideally we'd like to be able to auto update an array with all of the sample folders each time we run the overarching model. Is it possible to populate and update an array with the additional sample folder names/locations each time we run the overarching model rather than manually adding them?