A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications?
In biomechanics, movement is typically recorded by tracking the trajectories of anatomical landmarks previously marked using passive instrumentation, which entails several inconveniences. To overcome these disadvantages, researchers are exploring different markerless methods, such as pose estimation...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/24/6/1923 |
| _version_ | 1797239293541875712 |
|---|---|
| author | Ana V. Ruescas-Nicolau Enrique Medina-Ripoll Helios de Rosario Joaquín Sanchiz Navarro Eduardo Parrilla María Carmen Juan Lizandra |
| author_facet | Ana V. Ruescas-Nicolau Enrique Medina-Ripoll Helios de Rosario Joaquín Sanchiz Navarro Eduardo Parrilla María Carmen Juan Lizandra |
| author_sort | Ana V. Ruescas-Nicolau |
| collection | DOAJ |
| description | In biomechanics, movement is typically recorded by tracking the trajectories of anatomical landmarks previously marked using passive instrumentation, which entails several inconveniences. To overcome these disadvantages, researchers are exploring different markerless methods, such as pose estimation networks, to capture movement with equivalent accuracy to marker-based photogrammetry. However, pose estimation models usually only provide joint centers, which are incomplete data for calculating joint angles in all anatomical axes. Recently, marker augmentation models based on deep learning have emerged. These models transform pose estimation data into complete anatomical data. Building on this concept, this study presents three marker augmentation models of varying complexity that were compared to a photogrammetry system. The errors in anatomical landmark positions and the derived joint angles were calculated, and a statistical analysis of the errors was performed to identify the factors that most influence their magnitude. The proposed Transformer model improved upon the errors reported in the literature, yielding position errors of less than 1.5 cm for anatomical landmarks and 4.4 degrees for all seven movements evaluated. Anthropometric data did not influence the errors, while anatomical landmarks and movement influenced position errors, and model, rotation axis, and movement influenced joint angle errors. |
| first_indexed | 2024-04-24T17:49:14Z |
| format | Article |
| id | doaj.art-4137f7c6195b4665a0148e17a2a7fba0 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-24T17:49:14Z |
| publishDate | 2024-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-4137f7c6195b4665a0148e17a2a7fba02024-03-27T14:04:09ZengMDPI AGSensors1424-82202024-03-01246192310.3390/s24061923A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications?Ana V. Ruescas-Nicolau0Enrique Medina-Ripoll1Helios de Rosario2Joaquín Sanchiz Navarro3Eduardo Parrilla4María Carmen Juan Lizandra5Instituto de Biomecánica-IBV, Universitat Politècnica de València, Edifici 9C, Camí de Vera s/n, 46022 Valencia, SpainInstituto de Biomecánica-IBV, Universitat Politècnica de València, Edifici 9C, Camí de Vera s/n, 46022 Valencia, SpainInstituto de Biomecánica-IBV, Universitat Politècnica de València, Edifici 9C, Camí de Vera s/n, 46022 Valencia, SpainInstituto de Biomecánica-IBV, Universitat Politècnica de València, Edifici 9C, Camí de Vera s/n, 46022 Valencia, SpainInstituto de Biomecánica-IBV, Universitat Politècnica de València, Edifici 9C, Camí de Vera s/n, 46022 Valencia, SpainInstituto Universitario de Automática e Informática Industrial, Universitat Politècnica de València, Edifici 1F, Camí de Vera, s/n, 46022 Valencia, SpainIn biomechanics, movement is typically recorded by tracking the trajectories of anatomical landmarks previously marked using passive instrumentation, which entails several inconveniences. To overcome these disadvantages, researchers are exploring different markerless methods, such as pose estimation networks, to capture movement with equivalent accuracy to marker-based photogrammetry. However, pose estimation models usually only provide joint centers, which are incomplete data for calculating joint angles in all anatomical axes. Recently, marker augmentation models based on deep learning have emerged. These models transform pose estimation data into complete anatomical data. Building on this concept, this study presents three marker augmentation models of varying complexity that were compared to a photogrammetry system. The errors in anatomical landmark positions and the derived joint angles were calculated, and a statistical analysis of the errors was performed to identify the factors that most influence their magnitude. The proposed Transformer model improved upon the errors reported in the literature, yielding position errors of less than 1.5 cm for anatomical landmarks and 4.4 degrees for all seven movements evaluated. Anthropometric data did not influence the errors, while anatomical landmarks and movement influenced position errors, and model, rotation axis, and movement influenced joint angle errors.https://www.mdpi.com/1424-8220/24/6/1923markerlessdeep learninganatomical landmarkhuman pose estimationbiomechanicskeypoint augmentation |
| spellingShingle | Ana V. Ruescas-Nicolau Enrique Medina-Ripoll Helios de Rosario Joaquín Sanchiz Navarro Eduardo Parrilla María Carmen Juan Lizandra A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? Sensors markerless deep learning anatomical landmark human pose estimation biomechanics keypoint augmentation |
| title | A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? |
| title_full | A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? |
| title_fullStr | A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? |
| title_full_unstemmed | A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? |
| title_short | A Deep Learning Model for Markerless Pose Estimation Based on Keypoint Augmentation: What Factors Influence Errors in Biomechanical Applications? |
| title_sort | deep learning model for markerless pose estimation based on keypoint augmentation what factors influence errors in biomechanical applications |
| topic | markerless deep learning anatomical landmark human pose estimation biomechanics keypoint augmentation |
| url | https://www.mdpi.com/1424-8220/24/6/1923 |
| work_keys_str_mv | AT anavruescasnicolau adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT enriquemedinaripoll adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT heliosderosario adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT joaquinsanchiznavarro adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT eduardoparrilla adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT mariacarmenjuanlizandra adeeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT anavruescasnicolau deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT enriquemedinaripoll deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT heliosderosario deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT joaquinsanchiznavarro deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT eduardoparrilla deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications AT mariacarmenjuanlizandra deeplearningmodelformarkerlessposeestimationbasedonkeypointaugmentationwhatfactorsinfluenceerrorsinbiomechanicalapplications |