Currently, the data fetcher relies on images/labels having been downloaded manually. For usability of this code by other users, and on other stations, the fetcher should download the data at the time of class initialization, if it is not present already.
Currently, the data fetcher relies on images/labels having been downloaded manually. For usability of this code by other users, and on other stations, the fetcher should download the data at the time of class initialization, if it is not present already.
Currently, the datasets that need to be supported are: Kaggle Pathology Slide dataset: https://www.kaggle.com/code/taylorkern/tk-efficientnet - For the classification text Cell-Nuclei-Segmentation: https://morioh.com/p/11231a697d59 - For the segmentation task
The loading should be extendable to other datasets.