Thanh-Khoi Nguyen, Thien-Phuc Tran, Minh-Triet Tran · 2026 · arXiv
DOI: 10.48550/arXiv.2610.31293 · open-access PDF
O 3D Gaussian Splatting semântico online (3DGS) permite a compreensão de cenas de vocabulário aberto a partir de observações de streaming. No entanto, os métodos existentes recuperam o contexto geométrico perdido de características semânticas derivadas de imagens usando uma rede 3D aprendida adicionalmente, incorrendo em treinamento adicional e tempo de execução. Observamos que tais evidências geométricas-semânticas já podem existir no histórico de observação do agente e podem ser recuperadas em vez de aprendidas. Com base neste conhecimento, introduzimos o RecallGS (RecallGS - Geometry-Grounded Recall for Online Semantic Gaussian Splatting), um mecanismo leve e livre de treinamento.
Machine translation (Meta M2M100); the English original follows.
Online semantic 3D Gaussian Splatting (3DGS) enables open-vocabulary scene understanding from streaming observations. However, existing methods recover geometric context missing from image-derived semantic features using an additionally learned 3D network, incurring extra training and runtime overhead. We observe that such geometric-semantic evidence may already exist in the agent's observation history and can be retrieved rather than learned. Based on this insight, we introduce RecallGS (Geometry-Grounded Recall for Online Semantic Gaussian Splatting), a lightweight, training-free mechanism t
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31293, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange