Kintinuous: Spatially Extended KinectFusion

By Thomas Whelan, Michael Kaess, Maurice F. Fallon, Hordur Johannsson, John J. Leonard, John McDonald Jr.. 2012

729 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes (2017, background): “There is a large variety of algorithms targeting this scenario [59, 88, 7, 62, 37, 89, 42, 9, 90, 38, 12].”
OctNetFusion: Learning Depth Fusion from Data (2017, methodology): “We thus estimated the camera poses using Kintinuous [51] and visually inspect all models to remove those for which the pose tracking failed.”
Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016, neutral): “Mesh models have been also used in [258] and [257].”
A benchmark for RGB-D visual odometry, 3D reconstruction and SLAM (2014, methodology): “We evaluate 5 different odometers; (i) DVO [13] which performs full frame alignment between pairs of RGB-D frames making use of a robust cost function to improve robustness to outliers; (ii) FOVIS [14] which uses FAST feature correspondences between pairs of RGB-D frames to resolve camera motion, re”
Dense planar SLAM (2014, methodology): “However, KinectFusion and related methods [27, 19, 13] use nonparametric representations which are both heavyweight in terms of computing resources and lack semantic description.”
An Overview to Visual Odometry and Visual SLAM: Applications to Mobile Robotics (2015, background): “Whelan et al. [113] outlined this problem and proposed an extension to KinectFusion that enabled the approach to map larger environments.”
NICE-SLAM: Neural Implicit Scalable Encoding for SLAM (2021, methodology): “BAD-SLAM [38], Kintinuous [53], ORBSLAM2 [22]) still outperform the methods based on implicit scene representations (iMAP [42] and ours).”
Range sensor and silhouette fusion for high-quality 3D Scanning (2015, methodology): “Variants on the recently proposed KinectFusion algorithm have become particularly popular in reconstruction due to the method’s ability to reconstruct objects in realtime while recovering surface details [22], [30].”
iMAP: Implicit Mapping and Positioning in Real-Time (2021, methodology): “We compare with surfel-based BAD-SLAM [24], TSDF fusion Kintinuous [38], and sparse ORB-SLAM2 [16], state-of-the-art SLAM systems.”
Object discovery in 3D scenes via shape analysis (2013, methodology): “With tools such as Kinect Fusion [1], Kintinuous [2], and Open Source alternatives in the Point Cloud Library [3], it is now possible to collect detailed 3D meshes of entire scenes in real-time.”
School of Electrical and Computer Engineering Oklahoma State University , Stillwater , OK 74078 , USA (methodology): “Our method is used to generate four dense 3D models from depth streams of the augmented ICL-NUIM dataset and is evaluated against Kintinuous [24], DVO SLAM [17], SUN3D SfM [26], Manhattan frame (MF) reconstruction [28], and offline robust reconstruction [4] methods.”
DART: Dense Articulated Real-Time Tracking (2014, methodology): “Our tracking framework uses a local gradient based approach to find the pose of the model which best explains the data points observed in a depth frame along with a prior based on previous data.”
Shading-based refinement on volumetric signed distance functions (2015, methodology): “A prominent example of this line of research is Kinect Fusion [Newcombe et al. 2011; Izadi et al. 2011] and its extensions [Roth and Vona 2012; Whelan et al. 2012; Chen et al. 2013; Nießner et al. 2013], which led to signiﬁcant impact in the computer graphics, vision, and HCI communities.”
Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-board MAV planning (2016, methodology): “There have been multiple extensions to overcome this shortcoming, including using a moving fixed-size TSDF volume and meshing voxels exiting this volume [10], using an octree-based voxel grid [11], and allocating blocks of fixed size on demand in a method called voxel hashing [12].”
Robust real-time visual odometry for dense RGB-D mapping (2013, methodology): “Since the release of KinectFusion in 2011 a number of derived works have followed, notably our own system Kintinuous [1], the commercial ReconstructMe product [12], the KinFu Large Scale open source PCL project [13] and the Moving Volume KinectFusion work of Roth and Vona [14].”
MaskFusion: Real-Time Recognition, Tracking and Reconstruction of Multiple Moving Objects (2018, background): “Subsequent work [38, 51] showed that the same principles are applicable to large scale environments by choosing appropriate data structures.”
SceneNN: A Scene Meshes Dataset with aNNotations (2016, background): “KinectFusion [19] and its moving volume extension [33, 23, 20] can be applied to reconstructing a scene in real time.”
GS-SLAM: Dense Visual SLAM with 3D Gaussian Splatting (2023, methodology): “KinectFusion [41] performs camera tracking by iterative-closest-point and updates the scene via TSDF-Fusion.”
Fusion++: Volumetric Object-Level SLAM (2018, background): “Unlike many dense reconstruction systems [24, 39, 43, 38, 3, 8] we make no attempt to keep a dense representation of the entire scene.”