scholarly journals Are Existing Monocular Computer Vision-Based 3D Motion Capture Approaches Ready for Deployment? A Methodological Study on the Example of Alpine Skiing

Sensors ◽  
2019 ◽  
Vol 19 (19) ◽  
pp. 4323 ◽  
Author(s):  
Ostrek ◽  
Rhodin ◽  
Fua ◽  
Müller ◽  
Spörri
Keyword(s):  
Computer Vision ◽  
Center Of Mass ◽  
Three Dimensional ◽  
Image Data ◽  
Alpine Skiing ◽  
Data Set ◽  
Accuracy And Precision ◽  
Alpine Skiers ◽  
Human Pose ◽  
In Parenthesis

In this study, we compared a monocular computer vision (MCV)-based approach with the golden standard for collecting kinematic data on ski tracks (i.e., video-based stereophotogrammetry) and assessed its deployment readiness for answering applied research questions in the context of alpine skiing. The investigated MCV-based approach predicted the three-dimensional human pose and ski orientation based on the image data from a single camera. The data set used for training and testing the underlying deep nets originated from a field experiment with six competitive alpine skiers. The normalized mean per joint position error of the MVC-based approach was found to be 0.08 ± 0.01m. Knee flexion showed an accuracy and precision (in parenthesis) of 0.4 ± 7.1° (7.2 ± 1.5°) for the outside leg, and −0.2 ± 5.0° (6.7 ± 1.1°) for the inside leg. For hip flexion, the corresponding values were −0.4 ± 6.1° (4.4° ± 1.5°) and −0.7 ± 4.7° (3.7 ± 1.0°), respectively. The accuracy and precision of skiing-related metrics were revealed to be 0.03 ± 0.01 m (0.01 ± 0.00 m) for relative center of mass position, −0.1 ± 3.8° (3.4 ± 0.9) for lean angle, 0.01 ± 0.03 m (0.02 ± 0.01 m) for center of mass to outside ankle distance, 0.01 ± 0.05 m (0.03 ± 0.01 m) for fore/aft position, and 0.00 ± 0.01 m2 (0.01 ± 0.00 m2) for drag area. Such magnitudes can be considered acceptable for detecting relevant differences in the context of alpine skiing.

Automated Region-Based Vehicle Conflict Detection Using Computer Vision Techniques

2015 ◽  
Vol 2528 (1) ◽  
pp. 49-59
Author(s):  
Jinling Li ◽  
Yuhao Liu ◽  
Ahmed Tageldin ◽  
Mohamed H. Zaki ◽  
Greg Mori ◽  
...  
Keyword(s):  
Computer Vision ◽  
Three Dimensional ◽  
Data Set ◽  
Time To Collision ◽  
Video Footage ◽  
Precise Calculation ◽  
Vehicle Trajectories ◽  
Traffic Conflict ◽  
Point Representation ◽  
Feature Based

An approach for vehicle conflict analysis based on three-dimensional (3-D) vehicle detection is presented. Techniques for quantitative conflict measurements often use a point trajectory representation for vehicles. More accurate conflict measurement can be facilitated with a region-based vehicle representation instead. This paper describes a computer vision approach for extracting vehicle trajectories from video sequences. The method relied on a fusion of background subtraction and feature-based tracking to provide a three-dimensional (3-D) cuboid representation of the vehicle. Standard conflict measures, including time to collision and postencroachment time, were computed with the use of the 3-D cuboid vehicle representations. The use of these conflict measures was demonstrated on a challenging data set of video footage. Results showed that the region-based representation could provide more precise calculation of traffic conflict indicators compared with approaches based on a point representation.

Functional neuronavigation with magnetoencephalography: outcome in 50 patients with lesions around the motor cortex

1999 ◽  
Vol 91 (1) ◽  
pp. 73-79 ◽  
Author(s):  
Oliver Ganslandt ◽  
Rudolf Fahlbusch ◽  
Christopher Nimsky ◽  
Helmut Kober ◽  
Martin Möller ◽  
...  
Keyword(s):  
Motor Cortex ◽  
Three Dimensional ◽  
Image Data ◽  
Data Set ◽  
Content Type ◽  
Promising Tool ◽  
Neuronavigation System ◽  
Functional Neuronavigation ◽  
Intraoperative Recording ◽  
Fine Print

Object. The authors conducted a study to evaluate the clinical outcome in 50 patients with lesions around the motor cortex who underwent surgery in which functional neuronavigation was performed.Methods. The sensorimotor cortex was identified in all patients with the use of magnetoencephalography (MEG). The MEG-source localizations were superimposed onto a three-dimensional magnetic resonance image and the image data set was implemented into a neuronavigation system. Based on this setup, the surgeon chose the best surgical strategy. During surgery, the pre- and postcentral gyri were identified by neuronavigation and, in addition, the central sulcus was localized using intraoperative recording of somatosensory evoked potentials. In all cases MEG localizations of the sensory or motor cortex were correct. In 30% of the patients preoperative paresis improved, in 66% no additional deficits occurred, and in only 4% (two patients) deterioration of neurological function occurred. In one of these patients the deterioration was not related to the procedure.Conclusions. The method of incorporating functional data into neuronavigation systems is a promising tool that can be used in more radical surgery to lessen morbidity around eloquent brain areas.

VALIDATION OF MRI-BASED MEASUREMENTS OF SUPRASPINATUS MORPHOLOGY

2003 ◽  
Vol 07 (01) ◽  
pp. 15-23
Author(s):  
Tomotaka Nakajima ◽  
Richard E. Hughes ◽  
Kai-Nan An
Keyword(s):  
Data Processing ◽  
Spin Echo ◽  
Three Dimensional ◽  
Image Data ◽  
Supraspinatus Tendon ◽  
Fast Spin Echo ◽  
Acquisition Time ◽  
Data Set ◽  
Tendon Length ◽  
Voxel Data

The goal of this study was to visualize the supraspinatus tendon three-dimensionally using fast spin-echo (FSE) MRI and validate the accuracy of measuring the dimensions of the supraspinatus tendon based on 3D reconstructed images. Nine cadaver shoulders (51–84 y/o, mean 70.0 y/o) were imaged at conventional T2-weighted spin-echo (CSE), gradient echo (GRE), and 3D T2-weighted FSE sequences. Each "object" of the supraspinatus muscle, tendon and scapula was three-dimensionally reconstructed using ANALYZE™ image data processing software. The FSE images revealed significantly higher contrast of the tendon and contrast-to-noise ratios of the fat-to-tendon and fat-to-muscle. The length of the anterior, middle, and posterior portions of the tendon were measured in two ways: (1) from the three-dimensional reconstructed images, and (2) directly from the cadaver specimen using calipers. No statistically significant differences were found between the ANALYZE™ and caliper measurements using a paired t-test for the anterior (p = 0.55), middle (p = 0.57) and posterior (p = 0.44) portions of the supraspinatus. The 3D FSE sequence exhibits higher spatial resolution, spends shorter acquisition time, and constructs a voxel data set. These advantages can prevent blurring artifacts when imaging the supraspinatus tendon of a human body. Tendon length measurements derived from three-dimensional reconstructions using ANALYZE™ were found to be good estimates of actual tendon length. Therefore, the combination of FSE sequence and 3D image data processing provides a method for noninvasive quantitative analysis of supraspinatus tendon morphology. The results lay the groundwork for future quantitative studies of cuff pathology.

scholarly journals Functional neuronavigation with magnetoencephalography: outcome in 50 patients with lesions around the motor cortex

1999 ◽  
Vol 6 (3) ◽  
pp. E5 ◽  
Author(s):  
Oliver Ganslandt ◽  
Rudolf Fahlbusch ◽  
Christopher Nimsky ◽  
Helmut Kober ◽  
Martin Möller ◽  
...  
Keyword(s):  
Motor Cortex ◽  
Functional Data ◽  
Radical Surgery ◽  
Three Dimensional ◽  
Image Data ◽  
Data Set ◽  
Promising Tool ◽  
Neuronavigation System ◽  
Functional Neuronavigation ◽  
Intraoperative Recording

The authors conducted a study to evaluate the clinical outcome in 50 patients with lesions around the motor cortex who underwent surgery in which functional neuronavigation was performed. The sensorimotor cortex was identified in all patients with the use of magnetoencephalography (MEG). The MEG-source localizations were superimposed onto a three-dimensional magnetic resonance image, and the image data set was then implemented into a neuronavigation system. Based on this setup, the surgeon chose the best surgical strategy. During surgery, the pre- and postcentral gyrus were identified by neuronavigation, and in addition, the central sulcus was localized using intraoperative recording of somatosensory evoked potentials. In all cases MEG localizations of the sensory or motor cortex were correct. In 30% of the patients preoperative paresis improved, in 66% no additional deficits occurred, and in only 4% (two patients) deterioration of neurological function occurred. In one of these patients the deterioration was not related to the method. The method of incorporating functional data into neuronavigation systems is a promising tool that can be used in more radical surgery to cause less morbidity around eloquent brain areas.

Improved Three-Dimensional (3D) Resolution of Electron Tomograms Using Robust Mathematical Data-Processing Techniques

2017 ◽  
Vol 23 (6) ◽  
pp. 1121-1129 ◽  
Author(s):  
Toby Sanders ◽  
Ilke Arslan
Keyword(s):  
Data Processing ◽  
Electron Tomography ◽  
Full Range ◽  
Center Of Mass ◽  
Three Dimensional ◽  
Pt Nanoparticles ◽  
Specimen Preparation ◽  
Reconstruction Algorithms ◽  
Data Set ◽  
Tilt Series

AbstractElectron tomography has become an essential tool for three-dimensional (3D) characterization of nanomaterials. In recent years, advances have been made in specimen preparation and mounting, acquisition geometries, and reconstruction algorithms. All of these components work together to optimize the resolution and clarity of an electron tomogram. However, one important component of the data-processing has received less attention: the 2D tilt series alignment. This is challenging for a number of reasons, namely because the nature of the data sets and the need to be coherently aligned over the full range of angles. An inaccurate alignment may be difficult to identify, yet can significantly limit the final 3D resolution. In this work, we present an improved center-of-mass alignment model that allows us to overcome discrepancies from unwanted objects that enter the imaging area throughout the tilt series. In particular, we develop an approach to overcome changes in the total mass upon rotation of the imaging area. We apply our approach to accurately recover small Pt nanoparticles embedded in a zeolite that may otherwise go undetected both in the 2D microscopy images and the 3D reconstruction. In addition to this, we highlight the particular effectiveness of the compressed sensing methods with this data set.

scholarly journals Bruno Berenguel - Omnidirectional Image Data-set for Computer Vision Applications

2020 ◽  
Vol 8 ◽  
Author(s):  
Bruno Berenguel
Keyword(s):  
Computer Vision ◽  
Image Data ◽  
Data Set ◽  
Omnidirectional Image ◽  
Computer Vision Applications

Póster presentado en la IX Jornada de Jóvenes Investigadores del I3A

scholarly journals A Three-Dimensional Parametric Biomechanical Rider Model for Multibody Applications

2020 ◽  
Vol 10 (13) ◽  
pp. 4509
Author(s):  
Matteo Bova ◽  
Matteo Massaro ◽  
Nicola Petrone
Keyword(s):  
Center Of Mass ◽  
Three Dimensional ◽  
The Body ◽  
Experimental Comparison ◽  
Minor Effect ◽  
Data Sets ◽  
Data Set ◽  
Multibody Model ◽  
Inertial Parameters ◽  
A Minor

Bicycles and motorcycles are characterized by large rider-to-vehicle mass ratios, thus making estimation of the rider’s inertia especially relevant. The total inertia can be derived from the body segment inertial properties (BSIP) which, in turn, can be obtained from the prediction/regression formulas available in the literature. Therefore, a parametric multibody three-dimensional rider model is devised, where the four most-used BSIP formulas (herein named Dempster, Reynolds-NASA, Zatsiorsky–DeLeva, and McConville–Young–Dumas, after their authors) are implemented. After an experimental comparison, the effects of the main posture parameters (i.e., torso inclination, knee distance, elbow distance, and rider height) are analyzed in three riding conditions (sport, touring, and scooter). It is found that the elbow distance has a minor effect on the location of the center of mass and moments of inertia, while the effect of the knee distance is on the same order magnitude as changing the BSIP data set. Torso inclination and rider height are the most relevant parameters. Tables with the coefficients necessary to populate the three-dimensional rider model with the four data sets considered are given. Typical inertial parameters of the whole rider are also given, as a reference for those not willing to implement the full multibody model.

scholarly journals Imaged based identification of colombian timbers using the xylotron: a proof of concept international partnership

2021 ◽  
Vol 24 (1) ◽  
pp. 5-16
Author(s):  
Rafael E. Arévalo B. ◽  
Esperanza N. Pulido R. ◽  
Juan F. Solórzano G. ◽  
Richard Soares ◽  
Flavio Ruffinatto ◽  
...  
Keyword(s):  
Computer Vision ◽  
Image Data ◽  
Field Testing ◽  
Wood Identification ◽  
Proof Of Concept ◽  
Illegal Logging ◽  
Data Set ◽  
Final Model ◽  
International Partnership ◽  
Initial Identification

Field deployable computer vision wood identification systems can play a key role in combating illegal logging in the real world. This work used 764 xylarium specimens from 84 taxa to develop an image data set to train a classifier to identify 14 commercial Colombian timbers. We imaged specimens from various xylaria outside Colombia, trained and evaluated an initial identification model, then collected additional images from a Colombian xylarium (BOFw), and incorporated those images to refine and produce a final model. The specimen classification accuracy of this final model was ~ 97%, demonstrating that including local specimens can augment the accuracy and reliability of the XyloTron system. Our study demonstrates the first deployable computer vision model for wood identification in Colombia, developed on a timescale of months rather than years by leveraging international cooperation. We conclude that field testing and advanced forensic and machine learning training are the next logical steps.

scholarly journals Deep Features for Training Support Vector Machines

2021 ◽  
Vol 7 (9) ◽  
pp. 177
Author(s):  
Loris Nanni ◽  
Stefano Ghidoni ◽  
Sheryl Brahnam
Keyword(s):  
Computer Vision ◽  
Support Vector Machines ◽  
Vision System ◽  
Image Data ◽  
Support Vector ◽  
Data Sets ◽  
Data Set ◽  
Training Support ◽  
Reduction Techniques ◽  
Vector Machines

Features play a crucial role in computer vision. Initially designed to detect salient elements by means of handcrafted algorithms, features now are often learned using different layers in convolutional neural networks (CNNs). This paper develops a generic computer vision system based on features extracted from trained CNNs. Multiple learned features are combined into a single structure to work on different image classification tasks. The proposed system was derived by testing several approaches for extracting features from the inner layers of CNNs and using them as inputs to support vector machines that are then combined by sum rule. Several dimensionality reduction techniques were tested for reducing the high dimensionality of the inner layers so that they can work with SVMs. The empirically derived generic vision system based on applying a discrete cosine transform (DCT) separately to each channel is shown to significantly boost the performance of standard CNNs across a large and diverse collection of image data sets. In addition, an ensemble of different topologies taking the same DCT approach and combined with global mean thresholding pooling obtained state-of-the-art results on a benchmark image virus data set.

scholarly journals Omnidirectional Image Data-Set for Computer Vision Applications

2020 ◽  
Vol 8 ◽  
Author(s):  
Bruno Berenguel-Baeta ◽  
Jesus Bermudez-Cameo ◽  
Jose Jesus Guerrero
Keyword(s):  
Neural Networks ◽  
Computer Vision ◽  
Image Data ◽  
Full Information ◽  
Data Set ◽  
Omnidirectional Image ◽  
Computer Vision Applications ◽  
Instance Segmentation

In this paper we present an image data-set of different omnidirectional systems. The images include full information of colour, depth, instance segmentation and room layout. This dataset aims to help in the training and test of different neural networks and development of computer vision algorithms.

Sign in / Sign up

Export Citation Format

Share Document