Official implementation of "Consistency driven Sequential Transformers Attention Model for Partially Observable Scenes" [CVPR'22] - View it on GitHub