FLaME: Fast Lightweight Mesh Estimation Using Variational Smoothing on Delaunay Graphs

Greene, W. Nicholas; Roy, Nicholas

doi:10.1109/iccv.2017.502

Cited by 26 publications

(37 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To put this into the context of robotic control algorithms, this is not typically a property of SLAM based localization systems where more keyframes are accumulated over time although there has been recent work to cap (Maddern et al, 2012) or at least cull (Mur-Artal et al, 2015) the number of keyframes accumulated. While visual SLAM implementations based around SURF or SIFT can take several hundred milliseconds to extract features from each frame (Bay et al, 2006), recent SLAM implementations such as FLaME (Greene and Roy, 2017) have been demonstrated running on autonomous quadrotors at much higher framerates than our current Infomax implementation can achieve on the Jetson TX1. However, not only was FLaME implemented on an Intel CPU which Biddulph et al (2018) found to be 5× faster than a Jetson TX1, but the performance of our Infomax algorithm could be significantly improved.…”

Section: Summary and Discussionmentioning

confidence: 99%

Insect-Inspired Visual Navigation On-Board an Autonomous Robot: Real-World Routes Encoded in a Single Layer Network

Knight¹,

Sakhapov²,

Domcsek³

et al. 2019

The 2019 Conference on Artificial Life

View full text Add to dashboard Cite

Insect-Inspired models of visual navigation, that operate by scanning for familiar views of the world, have been shown to be capable of robust route navigation in simulation. These familiarity-based navigation algorithms operate by training an artificial neural network (ANN) with views from a training route, so that it can then output a familiarity score for any new view. In this paper we show that such an algorithmwith all computation performed on a small low-power robot-is capable of delivering reliable direction information along real-world outdoor routes, even when scenes contain few local landmarks and have high-levels of noise (from variable lighting and terrain). Indeed, routes can be precisely recapitulated and we show that the required computation and storage does not increase with the number of training views. Thus the ANN provides a compact representation of the knowledge needed to traverse a route. In fact, rather than losing information, there are instances where the use of an ANN ameliorates the problems of sub optimal paths caused by tortuous training routes. Our results suggest the feasibility of familiarity-based navigation for long-range autonomous visual homing.

show abstract

Section: Summary and Discussionmentioning

confidence: 99%

Insect-Inspired Visual Navigation On-Board an Autonomous Robot: Real-World Routes Encoded in a Single Layer Network

Knight¹,

Sakhapov²,

Domcsek³

et al. 2019

The 2019 Conference on Artificial Life

View full text Add to dashboard Cite

show abstract

“…Greene and Roy 23 cast the problem of surface reconstruction as a non‐local variational optimization on a time‐varying Delaunay graph of the geometric model. The graph can be tuned to improve the speed and quality of the textured model.…”

Section: Related Workmentioning

confidence: 99%

“…The graph can be tuned to improve the speed and quality of the textured model. Like the work in Greene and Roy, 23 Ummenhofer and Brox 24 proposed a variational method for surface reconstruction from point clouds with scale information, which can recover dense and multiscale surface model from a billion points. Lazar et al 25 considered that existing methods have mainly focused on improving the geometric aspects of the reconstruction, and little attention has been paid to the topological properties.…”

Section: Related Workmentioning

confidence: 99%

Parallel surface reconstruction for large‐scale scenes in the wild

Cao

Gao

Jia

2021

Circuit Theory & Apps

View full text Add to dashboard Cite

Recovering a high‐quality texture model in a short time still a pursued goal in both computer vision and graphics communities. Thus, in the past decade, several methods have been proposed for fitting a texture model from the dense point clouds. However, these methods are computational intensity and also suffer from noise especially in outdoor. Moreover, with the popularization of Unmanned Aerial Vehicles (UAVs), it is getting easier to capture image data. While modern methods have much novelty, they may spend a long time on big image datasets. To accelerate the process of texture modeling, in this paper we present a parallel approach to fitting a texture model from the dense point clouds. The presented method makes use of the parallel computing technology and is implemented in parallel octree structure as well as parallel marching cubes. Finally, we conduct a comprehensive experiment on several benchmarking datasets and experimental results show that our method outperforms the state‐of‐the‐art methods and has also a 20 times acceleration.

show abstract

“…Meshing and Noise Removal: We first project the 3D points onto an image and apply Delaunay triangulation to generate triangular meshes. Then, We use NLTGV minimization proposed by Greene et al [23] to remove noise on the meshes. NLTGV minimization allows us to smooth the meshes, thereby retaining local surface structures, unlike typical mesh denoising methods such as Laplacian smoothing.…”

Section: Dense Reconstructionmentioning

confidence: 99%

VITAMIN-E: VIsual Tracking and MappINg With Extremely Dense Feature Points

Yokozuka

Oishi

Thompson³

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

In this paper, we propose a novel indirect monocular SLAM algorithm called "VITAMIN-E," which is highly accurate and robust as a result of tracking extremely dense feature points. Typical indirect methods have difficulty in reconstructing dense geometry because of their careful feature point selection for accurate matching. Unlike conventional methods, the proposed method processes an enormous number of feature points by tracking the local extrema of curvature informed by dominant flow estimation. Because this may lead to high computational cost during bundle adjustment, we propose a novel optimization technique, the "subspace Gauss-Newton method", that significantly improves the computational efficiency of bundle adjustment by partially updating the variables. We concurrently generate meshes from the reconstructed points and merge them for an entire 3D model. The experimental results on the SLAM benchmark dataset EuRoC demonstrated that the proposed method outperformed state-of-theart SLAM methods, such as DSO, ORB-SLAM, and LSD-SLAM, both in terms of accuracy and robustness in trajectory estimation. The proposed method simultaneously generated significantly detailed 3D geometry from the dense feature points in real time using only a CPU.

show abstract

FLaME: Fast Lightweight Mesh Estimation Using Variational Smoothing on Delaunay Graphs

Cited by 26 publications

References 21 publications

Insect-Inspired Visual Navigation On-Board an Autonomous Robot: Real-World Routes Encoded in a Single Layer Network

Insect-Inspired Visual Navigation On-Board an Autonomous Robot: Real-World Routes Encoded in a Single Layer Network

Parallel surface reconstruction for large‐scale scenes in the wild

VITAMIN-E: VIsual Tracking and MappINg With Extremely Dense Feature Points

Contact Info

Product

Resources

About