Bridging the Domain Gap: Self-Supervised 3D Scene Understanding with Foundation Models | Read Paper on Bytez