Rethinking Multi-Modal Point Cloud Completion: Query-Aware Gating Attention and Gramian Volume Alignment
Multi-modal point cloud completion aims to recover complete 3D geometric structures from partial observations by integrating auxiliary data. Although image-guided techniques are well-established, the potential of natural language as a source of high-level semantic cues remains under-explored. Therefore, effectively int...