Abstract:We aim to make learned point cloud compression deployable for low-latency streaming on mobile systems. While learned point cloud compression has shown strong coding efficiency, practical deployment on mobile platforms remains challenging because neural inference and entropy coding still incur substantial runtime overhead. This issue is critical for immersive 3D communication, where dense geometry must be delivered under tight end-to-end (E2E) latency and compute constraints. In this paper, we present LEAN-3D, a compute-aware point cloud codec for low-latency streaming. LEAN-3D designs a lightweight learned occupancy model at the shallow levels of a sparse occupancy hierarchy, where structural uncertainty is highest, and develops a lightweight deterministic coding scheme for the deep hierarchy tailored to the near-unary regime. We implement the complete encoder/decoder pipeline and evaluate it on an NVIDIA Jetson Orin Nano edge device and a desktop host. In addition, LEAN-3D addresses the decoding failures observed in cross-platform deployment of learned codecs. Such failures arise from numerical inconsistencies in lossless entropy decoding across heterogeneous platforms. Experiments show that LEAN-3D achieves 3-5x latency reduction across datasets, reduces total edge-side energy consumption by up to 5.1x, and delivers lower sustained E2E latency under bandwidth-limited streaming. These results bring learned point cloud compression closer to deployable mobile 3D streaming.
Abstract:Deep learning has demonstrated strong capability in compressing point clouds. Within this area, entropy modeling for lossless compression is widely investigated. However, most methods rely solely on parent orsibling contexts and level-wise autoregression, which suffers from decoding latency on the order of 10 to 100 seconds. We propose HINT, a method that integrates temporal and spatial correlation for sequential point cloud compression. Specifically, it first uses a two stage temporal feature extraction: (i) a parent-level existence map and (ii) a child-level neighborhood lookup in the previous frame. These cues are fused with the spatial features via elementwise addition and encoded with a group-wise strategy. Experimental results show that HINT achieves encoding and decoding time at 105 ms and 140 ms, respectively, equivalent to 49.6x and 21.6x acceleration in comparison with G-PCC, while achieving up to bit rate reduction of 43.6%, in addition, consistently outperforming over the strong spatial only baseline (RENO).
Abstract:This paper presents a multimodal framework that integrates touch signals (contact points and surface normals) into 3D Gaussian Splatting (3DGS). Our approach enhances scene reconstruction, particularly under challenging conditions like low lighting, limited camera viewpoints, and occlusions. Different from the visual-only method, the proposed approach incorporates spatially selective touch measurements to refine both the geometry and appearance of the 3D Gaussian representation. To guide the touch exploration, we introduce a two-stage sampling scheme that initially probes sparse regions and then concentrates on high-uncertainty boundaries identified from the reconstructed mesh. A geometric loss is proposed to ensure surface smoothness, resulting in improved geometry. Experimental results across diverse scenarios show consistent improvements in geometric accuracy. In the most challenging case with severe occlusion, the Chamfer Distance is reduced by over 15x, demonstrating the effectiveness of integrating touch cues into 3D Gaussian Splatting. Furthermore, our approach maintains a fully online pipeline, underscoring its feasibility in visually degraded environments.
Abstract:Large language models (LLMs) have demonstrated remarkable performance on a variety of natural language tasks based on just a few examples of natural language instructions, reducing the need for extensive feature engineering. However, most powerful LLMs are closed-source or limited in their capability for languages other than English. In this technical report, we present Baichuan 2, a series of large-scale multilingual language models containing 7 billion and 13 billion parameters, trained from scratch, on 2.6 trillion tokens. Baichuan 2 matches or outperforms other open-source models of similar size on public benchmarks like MMLU, CMMLU, GSM8K, and HumanEval. Furthermore, Baichuan 2 excels in vertical domains such as medicine and law. We will release all pre-training model checkpoints to benefit the research community in better understanding the training dynamics of Baichuan 2.