Text this: A Vision-Based Pose Estimation of a Non-Cooperative Target Based on a Self-Supervised Transformer Network