Skip to main content

Table 2 Overview of studies comparing markerless with conventional motion analysis systems

From: A Review of the Evolution of Vision-Based Motion Analysis and the Integration of Advanced Computer Vision Methods Towards Developing a Markerless System

Publication

Movement(s) analysed

Markerless system description

Procedure/system for comparison

Number of cameras

Outcome

Trewartha et al. [110]

Starjump, somersaults

Gen-locked video cameras (50 Hz), subject-specific model

Manual digitising (TARGET system)

3

RMS differences for three movements ranged from 10 mm and 30 mm for pelvis location and between 2° and 8° for body configuration angles.

Corazza et al. [93]

Walking

Visual hull construction and a priori subject-specific model

Virtual environment (Poser software)

16

RMS errors of hip, knee and ankle angles ranged from 2.0° (hip abduction/adduction) to 9.0 (ankle dorsi/plantar flexion)

Mündermann et al. [10]

Walking

Video cameras (75 Hz), visual hull construction and a priori subject-specific model

Qualisys (120 Hz)

8

Average knee joint angle deviation: 2.3° (sagittal plane) and 1.6° (frontal plane).

Corazza et al. [86]

Walking

Video cameras (120 Hz), visual hull construction and a priori subject-specific model

Qualisys (120 Hz)

8

Average deviations between joint (hip, knee, ankle, shoulder, elbow and wrist) centres: 15 mm mean absolute error (ranged from 9 to 19 mm)

Choppin and Wheat [72]

Reaching, throwing, jumping

Microsoft Kinect (30 Hz)

Motion Analysis Corporation (60 Hz)

1 Kinect, 12 optoelectronic

Flexion/extension and abduction/adduction of hip, knee, elbow and shoulder; shoulder plane and elevation studied. Maximum abduction error: 44.1° and 13.9°. Maximum flexion error: 36.2° and 19.5° (NITE and IPIsoft tracking algorithms, respectively)

Ceseracciu et al. [107]

Walking

BTS

SMART-D

(100 Hz)

BTS

SMART-D

(200 Hz)

8

Maximum RMS differences range: 11.0° (ankle dorsi/ plantar flexion) to 34.7° (hip internal/external rotation)

Sandau et al. [108]

Walking

Monochrome cameras (75 Hz), unconstrained articulated model fit to 3D point clouds (aided by full body patterned suit)

Ariel Performance Analysis System

8

RMS differences in lower limb 3D angles ranged between 1.8° (hip abduction/adduction) and 4.9° (hip internal/external rotation)

Ong et al. [109]

Walking and jogging

Point Grey cameras (25 Hz)

Motion Analysis Corporation (100 Hz)

2 markerless, 8 marker-based

RMS differences ranged from 0.2° (knee abduction/adduction of jogging) to 1.0° (ankle dorsi/plantar flexion of walking). Significant differences between markerless and marker-based for the ankle joint angles.

  1. RMS root mean square