← Back to all news
arXiv AI By Mohammad Sadegh Salehi, Alexander J. P. Perkins, Igor Pardo Maurell, Ashkan Dabbagh, Raymond Wong

AmaraSpatial-10K: A Spatially and Semantically Aligned 3D Dataset for Spatial Computing and Embodied AI

Read the original on arXiv AI →

AmaraSpatial-10K is a curated collection of 10,071 synthetic 3D assets that are metric‑scaled, deterministically anchored, and accompanied by PBR maps, collision hulls, reference images, and multi‑sentence metadata. The dataset is paired with an evaluation suite that includes a Scale Plausibility Score, Concept Density metric, anchor‑error auditing, and a CLIP coherence protocol, and it outperforms existing collections such as Objaverse in CLIP Recall@5, stability in Habitat‑Sim, and human preference in scene placement. The authors also demonstrate that richer descriptions drive the retrieval gains observed. whyItMatters":"The dataset and evaluation tools directly address common defects in large 3D asset collections, enabling more reliable use in embodied AI, robotics, and spatial computing applications."

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv AI
Sep 25

GeoRefer-Bench: A Benchmark from Referring Pixels to Verifiable Geospatial Reasoning

GeoRefer-Bench is a new benchmark for verifiable geospatial referring segmentation that evaluates whether models correctly resolve spatial relations in overhead imagery. Each query is expressed as an executable logical form over a metric scene graph, and predictions are scored with Exact Query Success (EQS), requiring an exact match to the query’s referent set. The dataset contains 700 UAV scenes, 26,217 instances, 142,796 spatial relations, 20,916 executable queries across five reasoning levels, and additional paraphrases, unanswerable queries, counterfactual pairs, and leakage‑controlled splits.

By Shuaishuai Cao, Min Huang, Meng Tang, Xuan Liu, Youjin Wang, Hui Lin
computer-visionbenchmarks
More like this →
arXiv Computer Vision
Oct 22

AOV-GS: Active Open-Vocabulary Semantic Mapping with 3D Gaussian Splatting

AOV-GS is an active 3D Gaussian mapping system that uses frozen foundation models (SAM and CLIP) to guide exploration toward semantically informative regions. During coarse exploration, it re‑ranks geometrically promising views based on CLIP mask‑embedding novelty from SAM instance segmentation, biasing the robot’s trajectory. The system fits a lightweight scene‑specific language field offline to the collected CLIP embeddings, enabling open‑vocabulary text‑query segmentation on novel views and improving language‑field mIoU by 1.3–1.8 percentage points over geometry‑only planning on Replica and Matterport3D.

By Rodion Shkokov, Tatiana Zemskova, Dmitry Yudin
ragcomputer-visionrobotics
More like this →