STAR: A Spatial-Topology Aware Routing Framework for Generalizable 3D Scene Understanding

• URL: http://arxiv.org/abs/2608.11699v1
• Date: Wed, 12 Aug 2026 06:21:59 GMT
• Title: STAR: A Spatial-Topology Aware Routing Framework for Generalizable 3D Scene Understanding
• Abstract summary: We propose STAR (Spatial-Topology Aware Routing Framework) to overcome the challenge of constructing a unified 3D scene understanding model.<n>Specifically, we introduce a multi-attribute self-supervised pre-training branch, covering topological and textural variations, to anchor cross-domain structural priors.<n>Building upon this, we design a domain-aware expert branch with two mechanisms: Domain-Spatial-Guided Routing (DSR), which captures local topological variations from spatial context, and Entropy-controlled Dynamic Allocation (EDA), which adjusts the number of activated experts according to routing uncertainty.
• Score: 10.259254902492978
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Constructing a unified 3D scene understanding model has long been hindered by the topological discrepancies across sensor modalities. While applying the Mixture-of-Experts (MoE) architecture is a flexible approach for multi-domain 3D understanding, we observe that conventional feature-only MoE routers may underrepresent local sampling topology under semantic supervision, making expert allocation difficult when semantic consistency coexists with geometric heterogeneity. To overcome this challenge, we propose STAR (Spatial-Topology Aware Routing Framework). Specifically, we introduce a multi-attribute self-supervised pre-training branch, covering topological and textural variations, to anchor cross-domain structural priors. Building upon this, we design a domain-aware expert branch with two mechanisms: Domain-Spatial-Guided Routing (DSR), which captures local topological variations from spatial context, and Entropy-controlled Dynamic Allocation (EDA), which adjusts the number of activated experts according to routing uncertainty. Together, these branches combine stable cross-domain representation learning with adaptive expert allocation. Extensive experiments across various tasks, encompassing both indoor and outdoor scenes, demonstrate the effectiveness of STAR. It achieves 80.1% mIoU on the ScanNet validation set and 77.2% mIoU on S3DIS, consistently improving over strong baselines. Code is available at our project page (https://xmw666.github.io/STAR/).

Related papers

• Open-Text Aerial Detection: A Unified Framework For Aerial Visual Grounding And Detection [19.500762008628254]
  Open-Vocabulary Aerial Detection (OVAD) and Remote Sensing Visual Grounding (RSVG) have emerged as two key paradigms for aerial scene understanding.<n>We propose OTA-Det, the first unified framework that bridges both paradigms into a cohesive architecture.
  arXiv  Detail & Related papers  (2026-02-08T05:29:01Z)
• Beyond Binary Classification: A Semi-supervised Approach to Generalized AI-generated Image Detection [1.189955933770711]
  A critical vulnerability in current forensics is the failure of detectors to achieve cross-generator generalization.<n>We propose a semi-supervised approach that enhances binary classification by discovering latent architectural patterns within the "fake" class.
  arXiv  Detail & Related papers  (2025-11-23T16:02:27Z)
• DoReMi: A Domain-Representation Mixture Framework for Generalizable 3D Understanding [10.259254902492978]
  DoReMi is a Mixture-of-Experts (MoE) framework that jointly models Domain-aware Experts branch and a unified Representation branch.<n>DoReMi achieves 80.1% mIoU on ScanNet Val and 77.2% mIoU on S3DIS, demonstrating competitive or superior performance compared to existing approaches.
  arXiv  Detail & Related papers  (2025-11-14T12:32:45Z)
• Topology-Aware Modeling for Unsupervised Simulation-to-Reality Point Cloud Recognition [63.55828203989405]
  We introduce a novel Topology-Aware Modeling (TAM) framework for Sim2Real UDA on object point clouds.<n>Our approach mitigates the domain gap by leveraging global spatial topology, characterized by low-level, high-frequency 3D structures.<n>We propose an advanced self-training strategy that combines cross-domain contrastive learning with self-training.
  arXiv  Detail & Related papers  (2025-06-26T11:53:59Z)
• CrossEarth: Geospatial Vision Foundation Model for Domain Generalizable Remote Sensing Semantic Segmentation [85.22910555280414]
  CrossEarth is a vision foundation model for semantic segmentation.<n>CrossEarth demonstrates strong cross-domain generalization through a specially designed data-level Earth-Style Injection pipeline.<n>For the semantic segmentation task, we have curated an RSDG benchmark comprising 32 cross-domain settings across various regions, spectral bands, platforms, and climates.
  arXiv  Detail & Related papers  (2024-10-30T01:22:37Z)
• Boosting Cross-Domain Point Classification via Distilling Relational Priors from 2D Transformers [59.0181939916084]
  Traditional 3D networks mainly focus on local geometric details and ignore the topological structure between local geometries. We propose a novel Priors Distillation (RPD) method to extract priors from the well-trained transformers on massive images. Experiments on the PointDA-10 and the Sim-to-Real datasets verify that the proposed method consistently achieves the state-of-the-art performance of UDA for point cloud classification.
  arXiv  Detail & Related papers  (2024-07-26T06:29:09Z)
• Decomposition-based Unsupervised Domain Adaptation for Remote Sensing Image Semantic Segmentation [30.606689882397223]
  Unsupervised domain adaptation (UDA) techniques are vital for semantic segmentation in geosciences. Most existing UDA methods, which focus on domain alignment at the high-level feature space, struggle to simultaneously retain local spatial details and global contextual semantics. A novel decomposition scheme is proposed to guide domain-invariant representation learning.
  arXiv  Detail & Related papers  (2024-04-06T07:13:49Z)
• AFAN: Augmented Feature Alignment Network for Cross-Domain Object Detection [90.18752912204778]
  Unsupervised domain adaptation for object detection is a challenging problem with many real-world applications. We propose a novel augmented feature alignment network (AFAN) which integrates intermediate domain image generation and domain-adversarial training. Our approach significantly outperforms the state-of-the-art methods on standard benchmarks for both similar and dissimilar domain adaptations.
  arXiv  Detail & Related papers  (2021-06-10T05:01:20Z)
• I^3Net: Implicit Instance-Invariant Network for Adapting One-Stage Object Detectors [64.93963042395976]
  Implicit Instance-Invariant Network (I3Net) is tailored for adapting one-stage detectors. I3Net implicitly learns instance-invariant features via exploiting the natural characteristics of deep features in different layers. Experiments reveal that I3Net exceeds the state-of-the-art performance on benchmark datasets.
  arXiv  Detail & Related papers  (2021-03-25T11:14:36Z)
• Domain Conditioned Adaptation Network [90.63261870610211]
  We propose a Domain Conditioned Adaptation Network (DCAN) to excite distinct convolutional channels with a domain conditioned channel attention mechanism. This is the first work to explore the domain-wise convolutional channel activation for deep DA networks.
  arXiv  Detail & Related papers  (2020-05-14T04:23:24Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.