We present SLNet, a lightweight backbone for 3D point cloud recognition designed to achieve strong performance without the computational cost of many recent attention, graph, and deep MLP based models. The model is built on two simple ideas: NAPE (Nonparametric Adaptive Point Embedding), which captures spatial structure using a combination of Gaussian RBF and cosine bases with input adaptive bandwidth and blending, and GMU (Geometric Modulation Unit), a per channel affine modulator that adds only 2D learnable parameters. These components are used within a four stage hierarchical encoder with FPS+kNN grouping, nonparametric normalization, and shared residual MLPs. In experiments, SLNet shows that a very small model can still remain highly competitive across several 3D recognition tasks. On ModelNet40, SLNet-S with 0.14M parameters and 0.31 GFLOPs achieves 93.64% overall accuracy, outperforming PointMLP-elite with 5x fewer parameters, while SLNet-M with 0.55M parameters and 1.22 GFLOPs reaches 93.92%, exceeding PointMLP with 24x fewer parameters. On ScanObjectNN, SLNet-M achieves 84.25% overall accuracy within 1.2 percentage points of PointMLP while using 28x fewer parameters. For large scale scene segmentation, SLNet-T extends the backbone with local Point Transformer attention and reaches 58.2% mIoU on S3DIS Area 5 with only 2.5M parameters, more than 17x fewer than Point Transformer V3. We also introduce NetScore+, which extends NetScore by incorporating latency and peak memory so that efficiency can be evaluated in a more deployment oriented way. Across multiple benchmarks and hardware settings, SLNet delivers a strong overall balance between accuracy and efficiency. Code is available at: https://github.com/m-saeid/SLNet.
Figures & tables
Fig. 1 : NetScore and NetScore + across four datasets and two hardware platforms (RTX 3090 and Jetson Orin Nano), spanning 12 metric–dataset–hardware axes. SLNet-S and SLNet-M consistently occupy the outermost positions, establishing Pareto-optimal deployability across all evaluated configurations.
Fig. 2 : SLNet architecture: a NAPE+GMU front end, a four-stage hierarchical encoder with FPS, parameter-free normalization, and lightweight residual refinement, and task-specific heads for classification and segmentation.
Variant
Params
GFLOPs
Front-end
Local op.
SLNet-S
0.14M
0.31
NAPE+GMU
Shared MLP
SLNet-M
0.55M
1.22
NAPE+GMU
Shared MLP
SLNet-T
2.5M
6.5
Lin. proj. (6ch)
Point Trans
TABLE I : SLNet variant comparison.
Fig. 3 : NAPE: raw 3D coordinates mapped via fused Gaussian RBF and cosine bases with data-driven adaptive bandwidth, yielding a parameter-free geometry encoding.
Fig. 4 : GMU: per-channel affine recalibration of the NAPE output with only 2D learnable scalars.
Device
Batch
Workers
Software stack
Memory
RTX 3090
32
6
Ubuntu 22.04, CUDA 11.8
25.3 GB
Jetson Orin Nano
4
4
JetPack 6.2.2, CUDA 12.6, Python 3.10
8 GB
TABLE IV : Hardware and software used for efficiency benchmarking.
Fig. 5 : Qualitative comparison of saliency on ModelNet40 samples: top row—DGCNN stage-2 EdgeConv gradient saliency; middle and bottom—NAPE gradient saliency (channel widths 16 and 32). NAPE yields sharper, semantically focused responses despite operating at the input layer.
Modern neural architectures for 3D point cloud processing contain both convolutional layers and attention blocks, but the best way to assemble them remains unclear. We analyse the role of different computational blocks in 3D point cloud networks and find an intuitive behaviour: convolution is adequate to extract low-level geometry at high-resolution in early layers, where attention is expensive without bringing any benefits; attention captures high-level semantics and context in low-resolution, deep layers more efficiently, where convolution inflates the parameter count. Guided by this design principle, we propose a new, improved 3D point cloud backbone that employs convolutions in early stages and switches to attention for deeper layers. To avoid the loss of spatial layout information when discarding redundant convolution layers, we introduce a novel, parameter-free 3D positional encoding, PointROPE. The resulting LitePT model has 3.6× fewer parameters, runs 2× faster, and uses 2× less memory than the state-of-the-art Point Transformer V3, but nonetheless matches or outperforms it on a range of tasks and datasets. Code and models are available at: https://github.com/prs-eth/LitePT.
Yuanwen Yue, Damien Robert, Jianyuan Wang +4
ETH Zurich · University of Oxford · University of Zurich
Scene-level point cloud understanding remains challenging due to diverse geometries, imbalanced category distributions, and highly varied spatial layouts. Existing methods improve object-level performance but rely on static network parameters during inference, limiting their adaptability to dynamic scene data. We propose PointTPA, a Test-time Parameter Adaptation framework that generates input-aware network parameters for scene-level point clouds. PointTPA adopts a Serialization-based Neighborhood Grouping (SNG) to form locally coherent patches and a Dynamic Parameter Projector (DPP) to produce patch-wise adaptive weights, enabling the backbone to adjust its behavior according to scene-specific variations while maintaining a low parameter overhead. Integrated into the PTv3 structure, PointTPA demonstrates strong parameter efficiency by introducing two lightweight modules of less than 2% of the backbone's parameters. Despite this minimal parameter overhead, PointTPA achieves 78.4% mIoU on ScanNet validation, surpassing existing parameter-efficient fine-tuning (PEFT) methods across multiple benchmarks, highlighting the efficacy of our test-time dynamic network parameter adaptation mechanism in enhancing 3D scene understanding. The code is available at https://github.com/H-EmbodVis/PointTPA.
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.
Laurenz Reichardt, Nikolas Ebert, Oliver Wasenmüller