FAC: 3D Representation Learning via Foreground Aware Feature Contrast | Read Paper on Bytez