LiZIP: An Auto-Regressive Compression Framework for LiDAR Point Clouds

Aditya Shibu; Kayvan Karim; Claudio Zito

cs.ROMar 24, 2026

LiZIP: An Auto-Regressive Compression Framework for LiDAR Point Clouds

Authors: Aditya Shibu, Kayvan Karim, Claudio Zito

Organizations: MACS, Heriot Watt University, Dubai, United Arab Emirates

Abstract

The massive volume of data generated by LiDAR sensors in autonomous vehicles creates a bottleneck for real-time processing and vehicle-to-everything (V2X) transmission. Existing lossless compression methods often force a trade-off: industry standard algorithms (e.g., LASzip) lack adaptability, while deep learning approaches suffer from prohibitive computational costs. This paper proposes LiZIP, a lightweight, near-lossless zero-drift compression framework based on neural predictive coding. By utilizing a compact Multi-Layer Perceptron (MLP) to predict point coordinates from local context, LiZIP efficiently encodes only the sparse residuals. We evaluate LiZIP on the NuScenes and Argoverse datasets, benchmarking against GZip, LASzip, and Google Draco (configured with 24-bit quantization to serve as a high-precision geometric baseline). Results demonstrate that LiZIP consistently achieves superior compression ratios across varying environments. The proposed system achieves a 7.5%-14.8% reduction in file size compared to the industry-standard LASzip and outperforms Google Draco by 8.8%-11.3% across diverse datasets. Furthermore, the system demonstrates generalization capabilities on the unseen Argoverse dataset without retraining. Against the general purpose GZip algorithm, LiZIP achieves a reduction of 38%-48%. This efficiency offers a distinct advantage for bandwidth constrained V2X applications and large scale cloud archival.

Figures & tables

Panel of Figure 1 — Fig. 1: System Architecture of the LiZIP Compressor. Raw points are sorted and quantized. Processing occurs in parallel blocks : the first $k$ points (Anchors) initialize the context, while the MLP predicts the rest. Both anchors and residuals are byte-shuffled and entropy coded.

Fig. 2: Distribution of prediction residuals by the chosen $k=3,H=256$ configuration after neural prediction. Residuals are sharply peaked around zero, which improves entropy coding efficiency.

Panel of Figure 3 — Fig. 3: The LiZIP Binary File Format.

Arch.		Metrics
$k$	$H$	Lat. (s)	Size (KB)	Err. (mm)
3	256	0.19	185.41	0.010
3	512	0.31	186.13	0.010
3	1024	1.06	185.42	0.010
5	256	0.18	186.17	0.010
5	512	0.33	184.88	0.010

TABLE I: Performance Comparison of LiZIP Neural Architectures (C++ Implementation, Average Total Latency per Frame)

Fig. 4: Ablation Study of LiZIP Compression Stages. The “Waterfall” chart highlights that the Neural Predictor (MLP) is the primary driver of compression (53.2% reduction), distinguishing LiZIP from simple quantization schemes.

Method	Enc	Dec	Size	vs.	Err
	(ms)	(ms)	(KB)	LAZ	(mm)
LiZIP (lzma)	118 $\pm$ 15	74 $\pm$ 12	185.4 $\pm$ 28	-7.5%	0.010
LiZIP (zlib)	42 $\pm$ 8	33 $\pm$ 5	198.2 $\pm$ 29	-1.1%	0.010
LASzip	18 $\pm$ 4	15 $\pm$ 3	200.5 $\pm$ 17	—	0.011
Draco	41 $\pm$ 9	23 $\pm$ 6	203.3 $\pm$ 20	+1.4%	0.033
GZip	65 $\pm$ 12	40 $\pm$ 10	355.9 $\pm$ 42	+77.5%	0.000

TABLE II: Compression Performance on NuScenes (100 Frames)

Fig. 5: File Size Compression Breakdown of LiZIP against Baselines on NuScenes Dataset. Linear growth indicates consistent per-frame compression performance across all methods.

Method	Enc	Dec	Size	vs.	Err
	(ms)	(ms)	(KB)	LAZ	(mm)
LiZIP (lzma)	255 $\pm$ 25	160 $\pm$ 20	602.3 $\pm$ 6	-14.8%	0.017
LiZIP (zlib)	107 $\pm$ 12	84 $\pm$ 10	625.8 $\pm$ 8	-11.4%	0.017
LASzip	28 $\pm$ 5	23 $\pm$ 4	706.5 $\pm$ 6	—	0.018
Draco	56 $\pm$ 8	31 $\pm$ 5	679.2 $\pm$ 6	-3.9%	0.070
GZip	38 $\pm$ 6	24 $\pm$ 4	973.5 $\pm$ 13	+37.8%	0.000

TABLE III: Compression Performance on Argoverse (100 Frames)

Fig. 6: File Size Compression Breakdown of LiZIP against Baselines on Argoverse Dataset. Constant per-frame compression indicates uniform point cloud density across frames in this dataset.

Framework	Target	Hardware	Approach	Perf.
VoxelContext-Net	Geometry	GPU (2080Ti)	Octree+DL	-43.7%
OctSqueeze	Geometry	GPU (2080Ti)	Octree+Ent.	-15.0%
RCPCC	Geometry	CPU (i7)	Lossy Fit.	$\sim$ 60 $\times$
MNeT	Attributes	GPU (3090)	Probabilistic	-8.4%
LiZIP (Ours)	Geometry	CPU (i7)	Neural Pred.	-14.8%

TABLE IV: Qualitative Comparison of Neural Compression Frameworks

Point Cloud Compression

Point cloud compression reduces the storage and transmission cost of 3D points while preserving geometry, appearance, and task-relevant detail.

See all 22 papers

Mar 26, 2026

Towards Practical Lossless Neural Compression for LiDAR Point Clouds

LiDAR point clouds are fundamental to various applications, yet the extreme sparsity of high-precision geometric details hinders efficient context modeling, thereby limiting the compression speed and performance of existing methods. To address this challenge, we propose a compact representation for efficient predictive lossless coding. Our framework comprises two lightweight modules. First, the Geometry Re-Densification Module iteratively densifies encoded sparse geometry, extracts features at a dense scale, and then sparsifies the features for predictive coding. This module avoids costly computation on highly sparse details while maintaining a lightweight prediction head. Second, the Cross-scale Feature Propagation Module leverages occupancy cues from multiple resolution levels to guide hierarchical feature propagation, enabling information sharing across scales and reducing redundant feature extraction. Additionally, we introduce an integer-only inference pipeline to enable bit-exact cross-platform consistency, which avoids the entropy-coding collapse observed in existing neural compression methods and further accelerates coding. Experiments demonstrate competitive compression performance at real-time speed. Code will be released upon acceptance. Code is available at https://github.com/pengpeng-yu/FastPCC.

Pengpeng Yu, Haoran Li, Runqing Jiang +4

Sun Yat-sen University, China · Pengcheng Laboratory, China

Point Cloud Compression

Ask PDF

May 14, 2025

Efficient LiDAR Reflectance Compression via Scanning Serialization

Reflectance attributes in LiDAR point clouds provide essential information for downstream tasks but remain underexplored in neural compression methods. To address this, we introduce SerLiC, a serialization-based neural compression framework to fully exploit the intrinsic characteristics of LiDAR reflectance. SerLiC first transforms 3D LiDAR point clouds into 1D sequences via scan-order serialization, offering a device-centric perspective for reflectance analysis. Each point is then tokenized into a contextual representation comprising its sensor scanning index, radial distance, and prior reflectance, for effective dependencies exploration. For efficient sequential modeling, Mamba is incorporated with a dual parallelization scheme, enabling simultaneous autoregressive dependency capture and fast processing. Extensive experiments demonstrate that SerLiC attains over 2x volume reduction against the original reflectance data, outperforming the state-of-the-art method by up to 22% reduction of compressed bits while using only 2% of its parameters. Moreover, a lightweight version of SerLiC achieves > 10 fps (frames per second) with just 111K parameters, which is attractive for real-world applications.

Jiahao Zhu, Kang You, Dandan Ding +1

School of Information Science and Technology, Hangzhou Normal University, Hangzhou, China · School of Electronic Science and Engineering, Nanjing University, Nanjing, China

Point Cloud Compression

Ask PDF

May 2, 2026

PACE: Post-Causal Entropy Modeling for Learned LiDAR Point Cloud Compression

LiDAR point cloud compression is vital for autonomous systems to handle massive data from high-resolution sensors. While learned entropy modeling built upon octree structures yields high compression gains, it faces two critical bottlenecks: 1) prohibitive latency, particularly during decoding, caused by causal, multi-stage context modeling; and 2) a rigid performance-latency trade-off, preventing a single model from adapting to varying constraints. These limitations stem from the tight coupling between the context aggregation backbone and probability prediction. To address this, we propose PACE, a new framework that reformulates ancestral context aggregation as a non-causal backbone and confines causality to a lightweight, stage-scalable predictor, eliminating repetitive backbone executions and reducing computational overhead. The predictor supports an arbitrary number of prediction stages, enabling seamless adaptation across diverse performance-latency trade-offs without reloading parameters. Experiments demonstrate that PACE sets a new state-of-the-art in compression efficiency, achieving notable BD-BR savings and reducing decoding latency by over 90% in autoregressive mode, making it attractive for practical applications.

Jiahao Zhu, Kang You, Dandan Ding +1

School of Information Science and Technology, Hangzhou Normal University, Hangzhou, China. · School of Electronic Science and Engineering, Nanjing University, Nanjing, China.

Point Cloud Compression Airborne Laser Scanning Autoregressive Decoding

Ask PDF

Abstract

Point Cloud Compression

Explore similar work

Towards Practical Lossless Neural Compression for LiDAR Point Clouds

Efficient LiDAR Reflectance Compression via Scanning Serialization

PACE: Post-Causal Entropy Modeling for Learned LiDAR Point Cloud Compression