Focus-Then-Decide: Segmentation-Assisted Reinforcement Learning

We boost the performance of reinforcement learning under visual perturbations by introducing a segmentation model.

Download paper here