This repository contains an experimental setup for two popular Active Learning approoaches in the remote sensing domain: Margin sampling based sample selection and Entropy based sample selection strategies. The dataset used for this experiment is the AVIRIS HS Indian Pines dataset which can be downloaded directly using the code in this repo. Finally, experimental results are saved in the repository as pickled files.
The second version of this experiment includes a CLI. Run the script active-learning/AL_T29SND/run_active_learning_iteration.py
as if you were running any typical Python script. The image below demonstrates the expected behavior of the CLI under different contexts.