arXiv:2609.15137v1 Announce Type: cross
Abstract: 3D Gaussian language fields provide an explicit, spatially grounded representation for 3D visual question answering (VQA), but their dense semantic f...
By Davit Soselia, Joseph JaJa, Amitabh Varshney
The paper introduces Linguistic Trajectory Encoding (LTE), a hybrid representation that compresses dynamic object motion histories using natural language descriptions, sparse spatial anchors, and visual anchors. LTE adapts compression to motion complexity by anchoring periods without reliable observations to the last seen location while preserving accuracy with geometric waypoints and linguistic descriptions. Evaluated on the newly constructed Spatial Memory Benchmark (SMB) from EgoLife multi‑day recordings, LTE achieves 45.3 % success in semantic trajectory retrieval and 48.7 % in long‑horizon object retrieval, outperforming prior structured‑memory and VLM baselines, and compresses trajectories 8.7×–26.1× with sub‑second query latency on 24‑hour video.
By Tianyidan Xie, Shenyi Wang, Qiang Tang, Mingjie Wang, Zhicheng Qiu, Xuanfu Li, Zhan Xu, Jian Yang, Lanjun Wang, Zili Yi
The paper introduces a training‑free, target‑aware refinement method for 3D Gaussian language maps distilled from vision‑language features. By blending the visual embedding within user‑specified target regions with the corresponding text embedding using a convex combination, the method improves target localization while preserving open‑vocabulary queryability. The authors derive the optimal blending coefficient analytically, avoid hyperparameters, and demonstrate consistent gains across three benchmarks, including a real‑world robot navigation demo.
By Sibaek Lee, Kyeongsu Kang, Seongbo Ha, Yewon Jeon, Joonyeol Choi, Daegeol Ko, Jaehyuk Hur, Taemin Choi, Minsu Kim, Hyungjoo Ahn, Wonsok Yoo, Hyeonwoo Yu
Feed-forward 3D Gaussian Splatting enables efficient novel-view synthesis without per-scene optimization, but most existing methods assume a fixed set of context views and process them jointly. This limits their applicability to online scenarios where calibrated views arrive sequentially and the scene must be updated causally.
SpatialMem is a memory-based system that fuses 3D geometry, semantics, and language into a single queryable representation. Starting from casually recorded egocentric RGB video, it reconstructs metrically scaled interior scenes, identifies structural 3D anchors (walls, doors, windows), and links open‑vocabulary object nodes to evidence patches, visual embeddings, and two‑layer descriptions at 3D coordinates, enabling spatial‑relation reasoning, language‑directed retrieval, and QA without special sensors. The system is evaluated at scale on public benchmarks, achieving competitive scores on VSI‑Bench, OpenEQA, and ScanNet, while maintaining a compact index of about 14 KB per scene.
By Xinyi Zheng, Yunze Liu, Chi-Hao Wu, Fan Zhang, Hao Zheng, Wenqi Zhou, Walterio W. Mayol-Cuevas, Junxiao Shen
V-LUMEN introduces a visual lookup memory for vision‑language models by discretizing visual features with a frozen codebook, creating spatially aware memory keys via hashing, and modulating them with text‑conditioned relevance signals. The retrieved embeddings are fused into a frozen Qwen3‑VL‑2B backbone through a gated residual pathway, yielding substantial improvements on five reasoning‑intensive benchmarks and zero‑shot transfer to three disjoint image datasets. At a matched parameter budget, reallocating capacity from LoRA or MoE adapters to V‑LUMEN preserves accuracy on two backbones, demonstrating that external visual memory offers an independent capacity axis.
whyItMatters":"The paper shows that adding a non‑parametric visual lookup memory can significantly boost vision‑language model performance without increasing per‑token compute, offering a new way to scale such models."
By Miso Choi, Daekeun Kim, Eunji Kim, Jungbeom Lee