This article reviews the MPCT framework—a multiscale point‑cloud transformer built on a residual network that leverages permutation‑invariant self‑attention, point‑enhancement, and hierarchical feature aggregation to achieve state‑of‑the‑art results on ModelNet40 and ScanObjectNN datasets.
SIF3D is a scene‑aware 3D human motion forecasting framework that fuses observed motion, 3D point‑cloud scenes, and gaze through novel ternary intention‑aware and semantic‑coherence‑aware attention mechanisms, encoding with PointNet++ and Transformers, and decoding with a graph‑convolutional network, achieving state‑of‑the‑art results on GIMO and GTA‑1M benchmarks.
This article presents a Python project that reads point‑cloud data from CSV files, applies DBSCAN and KMeans clustering, extracts cluster features, and uses the Hungarian algorithm to match clusters across frames for tracking, complete with full source code and result visualization.
IPNet introduces a hybrid Image‑Point Cloud architecture that first extracts 2D visual features with a CNN, projects them into 3D point‑cloud space, and iteratively refines hand pose using a sparse‑anchor “aggregate‑interact‑propagate” scheme, achieving state‑of‑the‑art results on challenging hand‑object datasets.
This article introduces eight Python libraries—Open3D, Trimesh, PyVista, Vedo, Pyrender, PlotOptiX, Polyscope, and Simple‑3dviz—detailing their installation, core APIs, and step‑by‑step code examples for visualizing meshes, point clouds, and multidimensional data.
This tutorial introduces the Open3D library, shows how to install it, convert depth maps to point clouds, visualize facial point clouds, perform point‑cloud segmentation and clustering, and apply surface reconstruction methods such as Poisson, alpha shapes, and ball‑pivoting using Python code.
The Eagle.gl Web 3D engine, built atop Three.js, provides a unified 2D/3D GIS platform that efficiently loads, renders, and edits massive point‑cloud, vector, and model map data with real‑time LOD, GPU picking, DEM terrain integration, and configurable styling, supporting AMap’s HD‑map production and future autonomous‑driving simulation.
The article describes how high‑frequency LiDAR scanners on precision mapping vehicles suffer from Multi‑Time‑Around (MTA) errors—mis‑assigning distant returns to near ranges—and explains both internal laser strategies (continuity assumption and variable‑period emission) and a four‑step neighborhood‑weighting algorithm that reliably corrects these artifacts, restoring accurate point‑clouds for automated map generation.
This article explains the principles, components, generation process, and challenges of high‑precision topological and point‑cloud maps, and describes satellite‑based, map‑based, and fused high‑precision localization methods that underpin perception, prediction, planning, and control in autonomous driving systems.
With 5G accelerating video traffic, immersive media relies on advanced 3D representation, efficient point‑cloud and video‑based compression, standardized containers, and selective streaming to enable realistic VR/AR experiences across gaming, autonomous driving, and broadcasting, while Tencent’s lab drives standards and future XR applications.
