RANSAC
PulseAugur coverage of RANSAC — every cluster mentioning RANSAC across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New MMFE system unifies diverse 2D indoor representations for AI tasks
Researchers have developed the Multimodal Floorplan Encoder (MMFE), a system designed to process diverse 2D indoor representations like CAD drawings, raster images, and density maps into a unified latent grid. This appr…
-
New framework uses YOLO and RT-DETR for depth-aware pothole detection
Researchers have developed a depth-aware pothole detection framework that utilizes RGB-D sensors for improved accuracy. The study compared five different architectures, including YOLOv8n, YOLOv8nSeg, YOLOv9t, RTDETRL, a…
-
New method enhances LiDAR-camera calibration accuracy
Researchers have developed a novel method for accurately measuring 3D and 2D circular centers, crucial for tasks involving spatial sensors like LiDAR and cameras. This technique addresses conventional pipeline biases by…
-
New SfM method prunes view graphs for robust 3D reconstruction
Researchers have developed a new framework for robust global Structure-from-Motion (SfM) that addresses the sensitivity of existing methods to erroneous edges in the view graph. This novel approach partitions the view g…
-
New methods improve absolute pose estimation using visual-inertial fusion
Researchers have developed new methods for estimating the absolute pose of a device by fusing visual and inertial data, leveraging geometric information from feature descriptors like SIFT. These novel solvers, UP1PfAC a…
-
Structure from motion reconstructs 3D scenes from photos
Structure from motion (SfM) is a technique used to reconstruct 3D scenes from a collection of 2D photographs. The process involves several key steps: detecting repeatable interest points in images, matching these points…
-
New algorithm offers faster, precise camera pose estimation
Researchers have developed a new iterative method for estimating the relative pose between two cameras, which is a crucial step in Structure and Motion methods. This new algorithm, based on Powell's Dog Leg algorithm, o…
-
Unsupervised point cloud registration method uses self-distillation
Researchers have developed a novel unsupervised method for point cloud registration, a crucial task in robotics and autonomous driving. This self-distillation approach trains a student network using augmented views of p…
-
New frameworks tackle scale mismatch and rotation in point-cloud registration
Two new research papers introduce novel frameworks for point-cloud registration, a critical task in 3D perception for robotics. The first, R-SLPR, addresses the challenge of aligning small or incomplete point clouds wit…
-
New system enables GPS-free aerial geo-localization using satellite imagery
Researchers have developed a new system called NGPS (Next-Generation Positioning System) for high-altitude unmanned aerial vehicles (UAVs) that enables GPS-free absolute positioning. The system achieves this by matching…
-
AI model enhances microgravity combustion flame diameter measurement
Researchers have developed a new AI-driven method for accurately measuring the diameter of droplet flames in microgravity combustion images. This approach integrates the Segment Anything Model 2 (SAM2) with automatic pr…
-
New SemCom framework boosts real-time 3D reconstruction accuracy
Researchers have developed a novel Semantic Communication (SemCom) framework tailored for real-time mobile 3D reconstruction. This framework addresses the challenges of transmitting data from moving platforms to servers…
-
New DS-SAC framework offers deterministic geometric model estimation
Researchers have introduced DS-SAC, a novel deterministic framework for robust geometric model estimation in computer vision. Unlike traditional methods like RANSAC that rely on stochastic sampling, DS-SAC employs a den…
-
New RANSAC Score Eliminates User Parameters, Boosts Accuracy
Researchers have developed a novel RANSAC scoring method that eliminates the need for user-supplied parameters related to inlier scale. This new approach marginalizes the inlier scale analytically, allowing the score to…
-
New AI framework generates animal art from silhouettes
Researchers have developed Visual Retrieval-Augmented Generation (Visual-RAG), a new framework designed to computationally replicate human creativity in interpreting ambiguous shapes. The system generates animal art fro…
-
New solvers cut pose estimation complexity for multi-camera systems
Researchers have developed new, efficient minimal solvers for estimating the relative poses of multi-camera systems, crucial for applications like autonomous driving and robotics. These methods significantly reduce the …
-
New AI method improves bone angle estimation in medical imaging
Researchers have developed a novel method for robustly estimating bone angles in medical images, crucial for diagnosis and treatment. The approach combines a learning-based point candidate proposal with robust fitting t…
-
New methods improve point cloud registration accuracy and efficiency
Researchers have developed a new point cloud registration algorithm that uses probabilistic self-updating local correspondences and line vector sets to improve accuracy and efficiency. This method employs a dual RANSAC …
-
BIMStruct3D pipeline automates building information model generation from 3D scans
Researchers have developed BIMStruct3D, a novel pipeline that automates the creation of Building Information Models (BIM) from 3D point cloud data. This hybrid approach integrates learning-based semantic segmentation wi…
-
NONSAC framework offers scalable, robust model estimation for large datasets
Researchers have developed NONSAC, a novel framework designed for robust and scalable model estimation from extremely large datasets that contain noise and outliers. This method involves sampling non-minimal data subset…