RPCL:Open-Vocabulary 3D Semantic Segmentation via Region-level and Point-level Cross-modal Contrastive Learning
