SingleS2R: Single sample driven Sim-to-Real transfer for Multi-Source Visual-Tactile Information Understanding using multi-scale vision transformers.