Because the cost of inertial sensors is expensive and the intrusive wearing method is inconvenient, functional movement screening systems based on IMUs are not friendly to home-based or broad applications. The automatic scoring accuracy of this system can reach 66% 91% using the full movement feature set. 19 is an automatic functional movement screening system with 11 IMU sensors. Similarly, the study in 2020 by Wu et al. In 2020, the same team 18 used machine learning methods to train athletes’ movement data captured by IMU, and their model accuracy reached 75.1% to 84.7%, which is a huge improvement. However, the data for this study were captured by an optical motion capture system, which is expensive to apply. The accuracy of the model ranged from 70.66% to 92.91%. 17 used a data-driven method based on principal component analysis(PCA) and linear discriminant analysis to classify the performance of 542 athletes in seven dynamic screening movements (i.e., bird-dog, drop-jump, T-balance, step-down, L-lop, hop-down, and lunge). Based on the application of a self-set kinematics threshold for FMS scoring, discrepancies can be observed between automatic FMS scoring and manual scoring by professionals. 16 compared FMS scores rated by a certified FMS professional with an inertial-based motion capture system consisting of 17 inertial measurement unit (IMU) full-body sensors. With the rapid advance of sensing elements and computer technologies, some researchers have begun to focus on the direction of intelligent functional movement screens with the help of automatic action evaluation. In recent years, many application studies have argued that the higher the FMS score is, the lower the risk of sports injury, and individuals with low scores have a significantly increased risk of sports injury. The reliability and validity of FMS have been discussed for a long time, and numerous studies have confirmed that FMS is a reliable screening tool for physical functional evaluation 15. For FMS testing, seven fundamental movements (i.e., deep squat, hurdle step, in-line lunge, shoulder mobility, active straight raise, trunk stability push-up and rotary stability) 13, 14 are used to find the defects and deficiencies of the human body in terms of basic flexibility and stability.įMS research has attracted wide attention in recent years and provided important theoretical and practical perspectives. FMS can comprehensively evaluate the functional performance of individuals according to physical function evaluation criteria. It is a simple, efficient screening tool that predicts the risk of sports injury. Based on a literature review and the advice of exercise rehabilitation experts, a functional movement screen (FMS) is the most appropriate functional test and was proposed by Gray Cook in the 1990s 12. Considering that the target population of our experiment is the general population, the basic functional abilities that they need to perform is walking, running, and jumping in their daily life. Therefore, developing an automatic action evaluation system is very meaningful.īuilding an automatic action evaluation system is not an easy endeavor and involves choosing the most appropriate one from a wide diversity of movement evaluation systems. This process is inevitably influenced by their subjective attitude 11, and experienced experts are scarce in many areas. In most cases, experts, such as doctors, physiotherapists, and coaches, evaluate the individual’s physical state by observing specific limb movements based on their extensive experience. This dataset provides the opportunity for automatic action quality evaluation of FMS.Īction quality evaluation plays an important role in various fields: physical rehabilitation 1, 2, 3, 4, posture correction 5, sentiment analysis 6, 7, and sports training 8, 9, 10. Finally, our dataset contains a total of 1812 recordings, with 3624 episodes. The other is the multiview data collected from the two synchronized Azure Kinect sensors in front of and on the side of the subjects. One is the multimodal data provided, including color images, depth images, quaternions, 3D human skeleton joints and 2D pixel trajectories of 32 joints. The main strength of our database is twofold. Each episode was saved as one record and was annotated from 0 to 3 by three FMS experts. Specifically, shoulder mobility was performed only once by different subjects, while the other movements were repeated for three episodes each. This paper presents a dataset for vision-based autonomous Functional Movement Screen (FMS) collected from 45 human subjects of different ages (18–59 years old) executing the following movements: deep squat, hurdle step, in-line lunge, shoulder mobility, active straight raise, trunk stability push-up and rotary stability.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |