无需专用算子的高效3D点云模型，跨平台运行更快更省内存。
PointTransformerX: Portable and Efficient 3D Point Cloud Processing without Sparse Algorithms
- 纯PyTorch实现，用旋转位置编码直接建模3D空间关系。
- 参数少79.2%，速度提升1.6倍，内存仅需253MB，精度达原模型98.7%。
- 支持NVIDIA、AMD GPU及CPU，适合嵌入式和跨硬件部署。
3D点云感知长期依赖定制CUDA算子进行空间操作，限制了在非NVIDIA、AMD及嵌入式硬件上的可移植性与效率。本文提出PointTransformerX（PTX），一种全PyTorch原生的3D点云视觉变压器主干网络，完全移除自定义CUDA算子和外部库，同时保持竞争力的精度。PTX引入3D-GS-RoPE，一种旋转位置编码，直接在自注意力中编码3D空间关系，无需邻域构建；并以线性投影替代稀疏卷积块嵌入。通过推理时动态扩展注意力窗口，提升精度而无需重训练。结合重构的前馈网络，PTX在ScanNet上达到PointTransformer V3 98.7%的精度，参数减少79.2%，执行速度提升1.6倍，内存占用仅253 MB。PTX原生支持NVIDIA GPU、AMD GPU（ROCm）和CPU，为点云感知提供高效且可移植的基础。
原文摘要 · Abstract (English)
3D point cloud perception remains tightly coupled to custom CUDA operators for spatial operations, limiting portability and efficiency on non-NVIDIA, AMD, and embedded hardware. We introduce PointTransformerX (PTX), a fully PyTorch-native vision transformer backbone for 3D point clouds, removing all custom CUDA operators and external libraries while retaining competitive accuracy. PTX introduces 3D-GS-RoPE, a rotary positional embedding that encodes 3D spatial relationships directly in self-attention without neighborhood construction, and further replaces sparse convolutional patch embedding with a linear projection. PTX explores inference-time scaling of attention windows to improve accuracy without retraining. With a redesigned feed-forward network, PTX achieves 98.7\% of PointTransformer V3's accuracy on ScanNet with 79.2\% fewer parameters and executing 1.6\times faster while requiring just 253 MB memory. PTX runs natively on NVIDIA GPUs, AMD GPUs (ROCm), and CPUs, providing an efficient and portable foundation for point cloud perception.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。