SCANNET
PulseAugur coverage of SCANNET — every cluster mentioning SCANNET across labs, papers, and developer communities, ranked by signal.
10 day(s) with sentiment data
-
STAR framework enhances 3D scene understanding with novel routing techniques
Researchers have developed STAR, a novel framework designed to improve 3D scene understanding by addressing challenges posed by topological discrepancies across different sensor modalities. STAR utilizes a Mixture-of-Ex…
-
New PinpointQA benchmark tests MLLMs on indoor video spatial understanding
Researchers have introduced PinpointQA, a new benchmark designed to evaluate the spatial understanding capabilities of multimodal large language models (MLLMs) when processing indoor videos. The benchmark, built upon Sc…
-
New Self-Geometry pipeline enhances 3D vision model geometric consistency
Researchers have developed Self-Geometry, a novel test-time adaptation pipeline designed to enhance the geometric consistency of 3D vision foundation models. This method directly imposes explicit multi-view geometric co…
-
HKUST(GZ) unveils training-free 3D mapping system for robots
Researchers from The Hong Kong University of Science and Technology (Guangzhou) and Mohamed bin Zayed University of Artificial Intelligence have developed FreeOcc, a novel system for semantic occupancy prediction in 3D …
-
LLM Agents Generate Cinematic Camera Paths for 3D Scenes
Researchers have developed CinemaTraj, a novel framework that leverages LLM agents to generate cinematic camera trajectories for 3D scenes. This system takes a set of RGB-D images and a natural language prompt to decomp…
-
GeoStereo framework unifies stereo geometry estimation with diffusion priors
Researchers have introduced GeoStereo, a novel framework that unifies stereo geometry estimation for both disparity and surface normal prediction. This approach leverages diffusion priors to enhance performance in chall…
-
New method enhances 3D scene generation using latent space flow matching
Researchers have developed a new method called Latent Riemannian Flow Matching to improve 3D scene generation using geometric foundation models. This technique operates within the latent space of models like the Visual …
-
New methods advance panoptic segmentation with view synthesis and occlusion awareness · 2 sources tracked
Researchers are developing new methods for panoptic segmentation, a task that involves identifying and delineating every object instance and semantic region within an image. One approach leverages large view synthesis m…
-
New AI methods generate 3D scenes from multi-view images and sketches
Researchers have developed new methods for generating 3D scene assets from limited visual input. One approach, Scene-SAM3D, extends existing single-image 3D generation models to work with multiple calibrated views, impr…
-
New training-free pipeline advances 3D point-cloud segmentation
Researchers have developed a novel training-free pipeline for open-vocabulary 3D point-cloud segmentation. This method pairs a frozen 3D vision-language model, RegionPLC, with a frozen promptable concept segmenter, SAM3…
-
DINO-SLAM enhances neural representations in SLAM systems
Researchers have developed DINO-SLAM, a novel approach that integrates DINO features into Simultaneous Localization and Mapping (SLAM) systems. This method aims to improve both implicit (NeRF) and explicit (Gaussian Spl…
-
G2SR method offers fast, memory-efficient 3D surface reconstruction
Researchers have developed G2SR, a novel method for fast and memory-efficient 3D surface reconstruction using Gaussian-based techniques. Unlike existing end-to-end neural network approaches that demand significant compu…
-
AI system Holi-Spatial generates 3D annotations from 2D video without human input
A team of Chinese researchers has developed Holi-Spatial, an AI system that automatically generates 3D spatial data annotations from ordinary 2D videos without human intervention. This system bypasses the need for expen…
-
MAC-Splat framework enhances 3D reconstruction from sparse views
Researchers have introduced MAC-Splat, a novel training framework designed to improve the fidelity of 3D scene reconstruction from sparse camera views. This method addresses limitations in existing 3D Gaussian Splatting…
-
New WARM module enhances few-shot 3D point cloud segmentation
Researchers have developed a new method called the White Aggregation and Restoration Module (WARM) to improve few-shot 3D point cloud semantic segmentation. This technique addresses performance instability in existing m…
-
REMIND tracker achieves 90% IDF1 for indoor object re-identification
Researchers have developed REMIND, a novel online tracker designed for long-term re-identification of generic indoor objects using monocular RGB imagery. This system overcomes limitations of existing multi-object tracki…
-
REMIND system enhances indoor object re-identification with memory
Researchers have developed REMIND, a novel online tracking system designed for long-term re-identification of generic indoor objects using monocular RGB imagery. REMIND addresses challenges like significant viewpoint ch…
-
WanderDream dataset enables AI agents to reason via mental simulation
Researchers have introduced WanderDream, a novel dataset and framework designed to enable situated reasoning in AI agents through emulative simulation. This approach allows models to mentally explore future trajectories…
-
GEM-Occ framework enhances semantic occupancy mapping for indoor agents
Researchers have introduced GEM-Occ, a novel framework for semantic occupancy mapping in indoor environments. This system utilizes Gaussian Evidence Memory to represent occupied and free spaces, along with object semant…
-
New LMM enables metric-aware 3D spatial reasoning and grounding
Researchers have introduced Ground3D-LMM, a novel model designed to enhance natural language understanding of 3D environments. This model supports interactive conversations about 3D spaces by providing responses that ar…