arXiv:2511.20065v1 Announce Type: new 
Abstract: Point cloud compression methods jointly optimize bitrates and reconstruction distortion. However, balancing compression ratio and reconstruction quality is difficult because low-frequency and high-frequency components contribute differently at the same resolution. To address this, we propose FLaTEC, a frequency-aware compression model that enables the compression of a full scan with high compression ratios. Our approach introduces a frequency-aware mechanism that decouples low-frequency structures and high-frequency textures, while hybridizing latent triplanes as a compact proxy for point cloud. Specifically, we convert voxelized embeddings into triplane representations to reduce sparsity, computational cost, and storage requirements. We then devise a frequency-disentangling technique that extracts compact low-frequency content while collecting high-frequency details across scales. The decoupled low-frequency and high-frequency components are stored in binary format. During decoding, full-spectrum signals are progressively recovered via a modulation block. Additionally, to compensate for the loss of 3D correlation, we introduce an efficient frequency-based attention mechanism that fosters local connectivity and outputs arbitrary resolution points. Our method achieves state-of-the-art rate-distortion performance and outperforms the standard codecs by 78\% and 94\% in BD-rate on both SemanticKITTI and Ford datasets.

تم تقديم FLaTEC، وهو نموذج ضغط واعٍ للتردد، لتحسين كفاءة ضغط سحب النقاط LiDAR من خلال فصل الهياكل ذات التردد المنخفض عن القوام عالي التردد، مما يسمح بمعدلات ضغط عالية دون فقدان كبير في الجودة. يستخدم هذا النموذج ثلاثي الطائرات الكامنة كممثل مضغوط لتقليل التكاليف الحاسوبية ومتطلبات التخزين.

FLaTEC, un nuevo modelo de compresión consciente de la frecuencia, ha sido introducido para mejorar la eficiencia de la compresión de nubes de puntos LiDAR al desacoplar las estructuras de baja frecuencia de las texturas de alta frecuencia, permitiendo altas tasas de compresión sin una pérdida significativa de calidad. Este modelo utiliza triplanos latentes como representación compacta para minimizar los costos computacionales y los requisitos de almacenamiento.

FLaTEC, un nouveau modèle de compression conscient de la fréquence, a été introduit pour améliorer l'efficacité de la compression des nuages de points LiDAR en séparant les structures à basse fréquence des textures à haute fréquence, permettant ainsi des taux de compression élevés sans perte significative de qualité. Ce modèle utilise des triplans latents comme représentation compacte pour minimiser les coûts de calcul et les besoins de stockage.

FLaTEC, a new frequency-aware compression model, has been introduced to enhance the efficiency of LiDAR point cloud compression by decoupling low-frequency structures from high-frequency textures, allowing for high compression ratios without significant loss of quality. This model utilizes latent triplanes as a compact representation to minimize computational costs and storage needs.

FLaTEC: Frequency-Disentangled Latent Triplanes for Efficient Compression of LiDAR Point Clouds

arXiv:2502.09274v2 Announce Type: replace 
Abstract: 3D scene understanding is a critical yet challenging task in autonomous driving due to the irregularity and sparsity of LiDAR data, as well as the computational demands of processing large-scale point clouds. Recent methods leverage range-view representations to enhance efficiency, but they often adopt higher azimuth resolutions to mitigate information loss during spherical projection, where only the closest point is retained for each 2D grid. However, processing wide panoramic range-view images remains inefficient and may introduce additional distortions. Our empirical analysis shows that training with multiple range images, obtained from splitting the full point cloud, improves both segmentation accuracy and computational efficiency. However, this approach also poses new challenges of exacerbated class imbalance and increase in projection artifacts. To address these, we introduce FLARES, a novel training paradigm that incorporates two tailored data augmentation techniques and a specialized post-processing method designed for multi-range settings. Extensive experiments demonstrate that FLARES is highly generalizable across different architectures, yielding 2.1%~7.9% mIoU improvements on SemanticKITTI and 1.8%~3.9% mIoU on nuScenes, while delivering over 40% speed-up in inference.

تم تقديم نموذج تدريب جديد يسمى FLARES لتحسين تقسيم السيمانتك متعدد النطاقات باستخدام LiDAR، حيث يتناول التحديات المتعلقة بعدم انتظام وندرة بيانات LiDAR. يحسن هذا النهج دقة التقسيم وكفاءة الحوسبة من خلال التدريب باستخدام صور نطاق متعددة مستمدة من سحب النقاط الكاملة، على الرغم من أنه يقدم أيضًا تحديات جديدة مثل عدم التوازن في الفئات وعيوب الإسقاط.

Se ha introducido un nuevo paradigma de entrenamiento llamado FLARES para mejorar la segmentación semántica multi-rango LiDAR, abordando los desafíos relacionados con la irregularidad y escasez de datos LiDAR. Este enfoque mejora la precisión de la segmentación y la eficiencia computacional al entrenar con múltiples imágenes de rango derivadas de nubes de puntos completas, aunque también introduce nuevos desafíos como el desequilibrio de clases y los artefactos de proyección.

Un nouveau paradigme d'entraînement nommé FLARES a été introduit pour améliorer la segmentation sémantique multi-plage LiDAR, abordant les défis liés à l'irrégularité et à la rareté des données LiDAR. Cette approche améliore la précision de la segmentation et l'efficacité computationnelle en s'entraînant avec plusieurs images de portée dérivées de nuages de points complets, bien qu'elle introduise également de nouveaux défis tels que le déséquilibre de classe et les artefacts de projection.

A novel training paradigm named FLARES has been introduced to enhance LiDAR multi-range semantic segmentation, addressing challenges related to the irregularity and sparsity of LiDAR data. This approach improves segmentation accuracy and computational efficiency by training with multiple range images derived from full point clouds, although it also introduces new challenges such as class imbalance and projection artifacts.

FLaTEC: Frequency-Disentangled Latent Triplanes for Efficient Compression of LiDAR Point Clouds

Was this article worth reading? Share it

LucidQuery AI

Langfuse

flyline