Spatial-Then-Temporal Self-Supervised Learning for Video Correspondence | Read Paper on Bytez