The paper investigates how different 2D image backbones affect indoor semantic occupancy prediction in RGB‑D pipelines. By keeping the projection, depth branch, and occupancy head constant and swapping only the image encoder, the authors find that stronger backbones such as DINOv2 and BLIP2 significantly raise mIoU scores compared to the default ResNet‑50. These results show that the choice of image backbone is a major determinant of 3D occupancy accuracy, outweighing many specialized 3D modules.
By Shizhang Fanga, Wanling Yea, Qi Zheng
arXiv:2608. 07579v1 Announce Type: cross Abstract: The AI City Challenge 2026 Track 1 evaluates multi-camera 3D perception in large indoor warehouses under a synthetic-to-real (Sim2Real) setting; depth is available only for training and validation, so inference is RGB-only.
By Abdullah Naeem, Anav Katwal, Ayon Dey, Noman Khan, Md Tamjidul Hoque
O3N is a novel framework that performs open‑vocabulary occupancy prediction from a single omnidirectional RGB image. It introduces a polar‑spiral voxel embedding (PsM) for continuous 360° spatial representation, an Occupancy Cost Aggregation (OCA) module that unifies geometric and semantic supervision, and a Natural Modality Alignment (NMA) pathway that aligns visual, voxel, and text features. Experiments show state‑of‑the‑art results on QuadOcc and Human360Occ benchmarks, with strong cross‑scene generalization and semantic scalability.
By Mengfei Duan, Hao Shi, Fei Teng, Guoqiang Zhao, Yuheng Zhang, Zhiyong Li, Kailun Yang
VEOcc++ is a voxel‑centric framework that unifies online incremental semantic occupancy mapping with offline global map refinement by maintaining a persistent sparse voxel feature map. During exploration the map expands and integrates local features with reliability‑guided updates, while a context‑aware decoder produces the current semantic occupancy map. After exploration, cached local features are consolidated and combined with the online representation for multi‑scale spatial refinement, achieving state‑of‑the‑art IoU and mIoU scores on EmbodiedOcc‑ScanNet and demonstrating zero‑shot transfer to unseen environments on a mobile robot.
By Ruoyu Wang, Jiahan Li, Yong Liu, Kexin Zhang
The paper investigates how much semantic information is lost when frozen foundation models are combined for few‑shot 3D segmentation. By varying the number of retained semantic alternatives before fusion, the authors show that keeping the full distribution of class scores yields higher harmonic‑mean IoU than collapsing to a single class. Experiments on ScanNet200 and ScanNet++ confirm that full‑distribution fusion consistently outperforms top‑1 and other operators, and that most useful information is recovered by retaining a compact set of plausible alternatives.
By Silas Kwabla Gah, Ebenezer Owusu
Classical image correspondence is solved at the level of sparse keypoints or dense pixels, but the systems that consume these matches - object-level mapping, topological navigation, scene-graph maintenance - reason about whole objects. Recent work narrows this gap by matchng directly at the level of instance segments: a class-agnostic segmenter partitions each image, and per-segment descriptors are obtained by pooling features from large 3D foundation models over the masks.
EliGSiR is a continual RGB‑D mapping method that extends 3D Gaussian Splatting to online settings by adaptively allocating optimization resources. It introduces Map‑Guided View Scheduling to filter redundant views, Load‑Adaptive Fidelity to adjust supervision resolution, and Targeted Geometry Growth to add geometric capacity only where needed. The approach is evaluated on Replica, TUM RGB‑D, ScanNet++, and real sensor sequences, achieving higher reconstruction quality and faster performance than several baselines.
By Bj\"orn Ellensohn, Elmar Rueckert, Christian Rauch
The paper introduces DIFTA-3D, a method that replaces the task‑specific visual branch in IIFNet3D with a frozen DINOv3 foundation model for RGB‑D 3D instance detection. It employs a depth‑consistent feature pipeline that projects points into calibrated RGB‑D frames, filters features with a metric depth‑residual check, caches accepted DINOv3 features, and aggregates them within proposal‑aligned RoI grids. Extensive experiments on ScanNetV2 show that the DINOv3 control achieves mAP scores of 76.15/60.93 at IoU thresholds 0.25/0.50, while the Conservative VAID recipe improves these to 76.59/62.16, indicating a modest gain from the proposed transfer recipe.
By Linman Wang, ZiFei Zhang, Chunran Zheng, Xiwang Dong, Jiarong Lin
LangMap introduces a human‑verified benchmark for language‑conditioned goal navigation (LGN) that spans four hierarchical semantic levels—scene, room, region, and instance—within real‑world indoor 3D scans. The dataset, built on HM3D, contains 18K tasks with concise and detailed descriptions for 414 object categories, and its contrastive annotation protocol ensures high‑quality, discriminative region and instance labels. Evaluation shows that LangMap’s descriptions improve text‑to‑view matching accuracy by 23 points over GOAT‑Bench and achieve a 92.5% unique‑and‑correct match rate in an independent human audit, while a proposed RGB‑only baseline, PlaNaVid, attains top‑tier success rates without depth or 3D scene representations.
By Bo Miao, Weijia Liu, Jun Luo, Lachlan Shinnick, Jian Liu, Thomas Hamilton-Smith, Yuhe Yang, Zijie Wu, Vanja Videnovic, Feras Dayoub, Anton van den Hengel
EliGSiR is a continual RGB‑D mapping system that extends Gaussian splatting to handle online, bounded‑compute scenarios. It introduces Map‑Guided View Scheduling to filter redundant views, Load‑Adaptive Fidelity to adjust supervision resolution, and Targeted Geometry Growth to add structure only where needed. Experiments on Replica, TUM RGB‑D, ScanNet++ and real sensor data show that EliGSiR outperforms baselines in reconstruction quality while efficiently using the available mapping budget.
arXiv:2606.16569v2 Announce Type: replace
Abstract: Registering two captures of the same indoor space taken at different times underpins persistent spatial memory for robots and AR systems, yet the r...
By Zhiang Chen, Nahyuk Lee, Boyang Sun, Taein Kwon, Marc Pollefeys, Zuria Bauer, Sunghwan Hong
PanoBench is a new camera‑centric benchmark for panoptic occupancy prediction that unifies 3D geometry, semantics, and instance identities across synthetic and real‑world driving scenes. It combines three datasets—PanoBench‑CarlaOcc, PanoBench‑nuScenes, and PanoBench‑KITTI360—through canonical interfaces and unified evaluation protocols, providing multi‑resolution annotations with persistent instance identities. The benchmark introduces a novel reconstruction pipeline for real‑world data, a unified valid‑mask protocol, and analyses of observation fidelity, geometric completeness, spatial continuity, and temporal consistency, while also benchmarking methods on semantic, panoptic, and 4D occupancy forecasting.
whyItMatters":"PanoBench addresses the scarcity of instance‑aware annotations in panoptic occupancy research by offering comprehensive, high‑quality annotations and evaluation protocols that enable more robust development and comparison of autonomous driving perception methods."
By Yi Feng, Zizhan Guo, Junwu E, Yu Ma, Mengtan Zhang, Jin Wu, Wei Ye, Hanli Wang, Qijun Chen, Danwei Wang, Rui Fan