Dataset Information

A Video-Based Framework for Automatic 3D Localization of Multiple Basketball Players: A Combinatorial Optimization Approach.

ABSTRACT: Sports complexity must be investigated at competitions; therefore, non-invasive methods are essential. In this context, computer vision, image processing, and machine learning techniques can be useful in designing a non-invasive system for data acquisition that identifies players' positions in official basketball matches. Here, we propose and evaluate a novel video-based framework to perform automatic 3D localization of multiple basketball players. The introduced framework comprises two parts. The first stage is player detection, which aims to identify players' heads at the camera image level. This stage is based on background segmentation and on classification performed by an artificial neural network. The second stage is related to 3D reconstruction of the player positions from the images provided by the different cameras used in the acquisition. This task is tackled by formulating a constrained combinatorial optimization problem that minimizes the re-projection error while maximizing the number of detections in the formulated 3D localization problem.

SUBMITTER: Monezi LA

PROVIDER: S-EPMC7203428 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Video-Based Framework for Automatic 3D Localization of Multiple Basketball Players: A Combinatorial Optimization Approach.

Monezi Lucas Antônio LA Calderani Junior Anderson A Mercadante Luciano Allegretti LA Duarte Leonardo Tomazeli LT Misuta Milton S MS

Frontiers in bioengineering and biotechnology 20200430

Sports complexity must be investigated at competitions; therefore, non-invasive methods are essential. In this context, computer vision, image processing, and machine learning techniques can be useful in designing a non-invasive system for data acquisition that identifies players' positions in official basketball matches. Here, we propose and evaluate a novel video-based framework to perform automatic 3D localization of multiple basketball players. The introduced framework comprises two parts. T ...[more]

PMID: 32426334

Similar Datasets

Project description:Three-dimensional (3D) fetal neurosonography is used clinically to detect cerebral abnormalities and to assess growth in the developing brain. However, manual identification of key brain structures in 3D ultrasound images requires expertise to perform and even then is tedious. Inspired by how sonographers view and interact with volumes during real-time clinical scanning, we propose an efficient automatic method to simultaneously localize multiple brain structures in 3D fetal neurosonography. The proposed View-based Projection Networks (VP-Nets), uses three view-based Convolutional Neural Networks (CNNs), to simplify 3D localizations by directly predicting 2D projections of the key structures onto three anatomical views. While designed for efficient use of data and GPU memory, the proposed VP-Nets allows for full-resolution 3D prediction. We investigated parameters that influence the performance of VP-Nets, e.g. depth and number of feature channels. Moreover, we demonstrate that the model can pinpoint the structure in 3D space by visualizing the trained VP-Nets, despite only 2D supervision being provided for a single stream during training. For comparison, we implemented two other baseline solutions based on Random Forest and 3D U-Nets. In the reported experiments, VP-Nets consistently outperformed other methods on localization. To test the importance of loss function, two identical models are trained with binary corss-entropy and dice coefficient loss respectively. Our best VP-Net model achieved prediction center deviation: 1.8 ± 1.4 mm, size difference: 1.9 ± 1.5 mm, and 3D Intersection Over Union (IOU): 63.2 ± 14.7% when compared to the ground truth. To make the whole pipeline intervention free, we also implement a skull-stripping tool using 3D CNN, which achieves high segmentation accuracy. As a result, the proposed processing pipeline takes a raw ultrasound brain image as input, and output a skull-stripped image with five detected key brain structures.

Project description:ImportanceThere is a paucity of data detailing cardiac remodeling in female athletes compared with male athletes. The lack of reference cardiac data for elite female basketball players or female athletes of similar size makes it difficult to differentiate athletic remodeling from potential underlying cardiac disorders in this population of athletes.ObjectiveTo assess cardiac structure and function in elite female basketball players.Design, setting, and participantsThis cross-sectional echocardiographic study included 140 Women's National Basketball Association (WNBA) athletes on active rosters for the 2017 season. The WNBA mandates annual preseason stress echocardiograms for each athlete. The WNBA has partnered with Columbia University to annually perform a review of these studies. Data analysis was performed from June 7, 2017, to October 5, 2017.Main outcomes and measuresEchocardiographic variables included left ventricular (LV) dimensions, wall thickness, mass, prevalence of LV hypertrophy, aortic dimensions, right ventricular (RV) dimension, and right and left atrial size. Linear regression was used to assess the associations between cardiac structure and function with body size quantified as body surface area (BSA) in the primary analysis.ResultsA total of 140 female athletes (mean [SD] age, 26.8 [3.9] years; 105 [75.0%] African American) participated in the study. Mean (SD) athlete height was 183.4 (9.0) cm, and mean (SD) BSA was 2.02 (0.18) m2. Compared with guideline-defined normal values, LV enlargement was present in 36 athletes (26.0%) and 57 athletes (42.2%) had RV enlargement. There was a linear correlation between LV and RV cavity sizes and BSA extending to the uppermost biometrics (LV cavity size: r, 0.48; RV cavity size: r, 0.32; P < .001 for both). Maximal left ventricular wall thickness (LVWT) ranged from 0.6 to 1.4 cm, with 78 athletes (55.7%) having LVWT of 1.0 cm or greater and only 1 athlete (0.7%) having LVWT greater than 1.3 cm. Twenty-three athletes (16.4%) met the criteria for left ventricular hypertrophy (LVH) (>95 g/m2). Eccentric LVH was present in 16 athletes (69.6%), concentric LVH in 7 athletes (30.4%), and concentric remodeling in 27 athletes (19.3%). Mean aortic root diameter was 3.1 cm (95% CI, 3.0-3.2). Only 2 athletes (1.4%) had guideline-defined aortic enlargement compared with a range of 18% to 42% for left and right ventricular and atrial enlargement.Conclusions and relevanceIn this study, increased cardiac dimensions were frequently observed in WNBA athletes. Both BSA and physiologic remodeling affected cardiac morphologic findings. This study may provide a framework to define the range of athletic cardiac remodeling exhibited by elite female basketball players.

Project description:ObjectiveProfessional Basketball players are at high risk of Achilles tendon rupture. Despite this, there remains limited research into the factors affecting rehabilitation and the long term outcomes of these players. Our aim is to quantify the effect of a player's Achilles tendon rupture on their post-injury performance, and also to explore for correlations between their recovery timeline and pre-injury characteristics. Creation of an injury timeline of past incidents will allow injured players to better track their progress and also inform them about the probable impact on their careers.HypothesisPlayers with Achilles tendon rupture injury will exhibit decreased performance compared to their pre-injury self and their non-injured peers after recovery.MethodsProfessional basketball players who sustained a unilateral Achilles tendon rupture from 1992 to 2016 were collected. 12 players met our inclusion criteria and their Player Efficiency Ratings (PER) were obtained as primary outcome measures; matched controls were chosen based on the PER, Age and playing position. The players' index season PER was compared against the PER during the 10 games immediately following the players' return and the PER of their post-injury peak performing season. The same data analysis was performed against their control group. To investigate the factors affecting the recovery and long-term consequences of their injury, we correlated the variables of Age, BMI, Time of Injury and pre-injury PER with the player's time to return to play and their post-injury PER.Results2 out of 12 players failed to return to playing in the elite professional league following an Achilles Tendon Rupture, others returned after a mean recovery time of 10 months. When compared to players' index PER, the mean PER reduction during the 10 games immediately following the players' return was 7.15 (P < .000). Players on average took 1.8 seasons to reach their post-injury peak performance, with only 1 player returning to his pre-injury performance. Others suffered a mean PER reduction of 3.5 (P = .004) when compared to their index PER and 5.4 (P = .045) against their matched controls.ConclusionAchilles tendon rupture can be a career-ending injury for professional basketball players. They are expected to miss 10 months for rehabilitation and reach their post-injury peak performance level at the 2nd season back. The post-injury peak performance is significantly worse than the pre-injury level, but is similar to matched non-injured players.

Dataset Information

A Video-Based Framework for Automatic 3D Localization of Multiple Basketball Players: A Combinatorial Optimization Approach.

Publications

A Video-Based Framework for Automatic 3D Localization of Multiple Basketball Players: A Combinatorial Optimization Approach.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets