https://sam.ensam.eu:443 The DSpace digital repository system captures, stores, indexes, preserves, and distributes digital research material. Mon, 16 Mar 2026 02:56:33 GMT 2026-03-16T02:56:33Z http://hdl.handle.net/10985/20116 Evaluation of CNN-Based Human Pose Estimation for Body Segment Lengths Assessment VAFADAR, Saman; BOËSSÉ, Matthieu; SKALLI, Wafa; GAJNY, Laurent Human pose estimation (HPE) methods based on convolutional neural networks (CNN) have demonstrated significant progress and achieved state-of-the-art results on human pose datasets. In this study, we aimed to assess the perfor-mance of CNN-based HPE methods for measuring anthropometric data. A Vicon motion analysis system as the reference system and a stereo vision system recorded ten asymptomatic subjects standing in front of the stereo vision system in a static posture. Eight HPE methods estimated the 2D poses which were transformed to the 3D poses by using the stereo vision system. Percentage of correct keypoints, 3D error, and absolute error of the body segment lengths are the evaluation measures which were used to assess the results. Percentage of correct keypoints – the stand-ard metric for 2D pose estimation – showed that the HPE methods could estimate the 2D body joints with a minimum accuracy of 99%. Meanwhile, the average 3D error and absolute error for the body segment lengths are 5 cm. Tue, 01 Jan 2019 00:00:00 GMT http://hdl.handle.net/10985/20116 2019-01-01T00:00:00Z VAFADAR, Saman BOËSSÉ, Matthieu SKALLI, Wafa GAJNY, Laurent Human pose estimation (HPE) methods based on convolutional neural networks (CNN) have demonstrated significant progress and achieved state-of-the-art results on human pose datasets. In this study, we aimed to assess the perfor-mance of CNN-based HPE methods for measuring anthropometric data. A Vicon motion analysis system as the reference system and a stereo vision system recorded ten asymptomatic subjects standing in front of the stereo vision system in a static posture. Eight HPE methods estimated the 2D poses which were transformed to the 3D poses by using the stereo vision system. Percentage of correct keypoints, 3D error, and absolute error of the body segment lengths are the evaluation measures which were used to assess the results. Percentage of correct keypoints – the stand-ard metric for 2D pose estimation – showed that the HPE methods could estimate the 2D body joints with a minimum accuracy of 99%. Meanwhile, the average 3D error and absolute error for the body segment lengths are 5 cm. http://hdl.handle.net/10985/22659 Assessment of a novel deep learning-based marker-less motion capture system for gait study VAFADAR, Saman; SKALLI, Wafa; BONNET-LEBRUN, Aurore; ASSI, Ayman; GAJNY, Laurent Background. Marker-less systems based on digital video cameras and deep learning for gait analysis could have a deep impact in clinical routine. A recently developed system has shown promising results in terms of joint center position but has not been yet evaluated in terms of gait outcomes. Research question. How does this novel marker-less system compare to a marker-based reference system in terms of clinically relevant gait parameters? Methods. The deep learning method behind the developed marker-less system was trained on a dedicated dataset consisting of forty-one asymptomatic and pathological subjects each performing ten walking trials. The system could estimate the three-dimensional position of seventeen joint centers or keypoints (e.g., neck, shoulders, hip, knee, and ankles). We evaluated the marker-less system against a marker-based system in terms of differences in joint position (Euclidean distance), detection of gait events (e.g., heel strike and toe-off), spatiotemporal parameters (e.g., step length, time), kinematic parameters (e.g., hip and knee extension-flexion), and inter-trial reliability for kinematic parameters. Results. The marker-less system was able to estimate the three-dimensional position of joint centers with a mean difference of 13.1 mm (SD = 10.2 mm). 99% of the estimated gait events were estimated within 10 milliseconds of the corresponding reference values. Estimated spatiotemporal parameters showed zero bias. The mean and standard deviation of the differences of the estimated kinematic parameters varied by parameter (for example, the mean and standard deviation for knee extension flexion angle were -3.0° and 2.7°). Inter-trial reliability of the measured parameters was similar to that of the marker-based references. Significance. The developed marker-less system can measure the spatiotemporal parameters within the range of the minimum detectable changes obtained using the marker-based reference system. Moreover, except for hip extension flexion, the system showed promising results in terms of several kinematic parameters. Sun, 01 May 2022 00:00:00 GMT http://hdl.handle.net/10985/22659 2022-05-01T00:00:00Z VAFADAR, Saman SKALLI, Wafa BONNET-LEBRUN, Aurore ASSI, Ayman GAJNY, Laurent Background. Marker-less systems based on digital video cameras and deep learning for gait analysis could have a deep impact in clinical routine. A recently developed system has shown promising results in terms of joint center position but has not been yet evaluated in terms of gait outcomes. Research question. How does this novel marker-less system compare to a marker-based reference system in terms of clinically relevant gait parameters? Methods. The deep learning method behind the developed marker-less system was trained on a dedicated dataset consisting of forty-one asymptomatic and pathological subjects each performing ten walking trials. The system could estimate the three-dimensional position of seventeen joint centers or keypoints (e.g., neck, shoulders, hip, knee, and ankles). We evaluated the marker-less system against a marker-based system in terms of differences in joint position (Euclidean distance), detection of gait events (e.g., heel strike and toe-off), spatiotemporal parameters (e.g., step length, time), kinematic parameters (e.g., hip and knee extension-flexion), and inter-trial reliability for kinematic parameters. Results. The marker-less system was able to estimate the three-dimensional position of joint centers with a mean difference of 13.1 mm (SD = 10.2 mm). 99% of the estimated gait events were estimated within 10 milliseconds of the corresponding reference values. Estimated spatiotemporal parameters showed zero bias. The mean and standard deviation of the differences of the estimated kinematic parameters varied by parameter (for example, the mean and standard deviation for knee extension flexion angle were -3.0° and 2.7°). Inter-trial reliability of the measured parameters was similar to that of the marker-based references. Significance. The developed marker-less system can measure the spatiotemporal parameters within the range of the minimum detectable changes obtained using the marker-based reference system. Moreover, except for hip extension flexion, the system showed promising results in terms of several kinematic parameters. http://hdl.handle.net/10985/25103 Full-body Postural Alignment Analysis Through Barycentremetry KHALIFE, Marc; VERGARI, Claudio; ASSI, Ayman; GUIGUI, Pierre; ATTALI, Valerie; VALENTIN, Rémi; VAFADAR, Saman; FERRERO, Emmanuelle; SKALLI, Wafa Study design: Multicentric retrospective. Objective: The study of center of mass (COM) locations (i.e. barycentremetry) can help us understand postural alignment. This study goal was to determine relationships between COM locations and global postural alignment X-ray parameters in healthy subjects. The second objective was to determine the impact on spinopelvic alignment of increased distance between anterior body envelope and spine at lumbar apex level. Summary of background data: Unexplored relationship between COM location and spinopelvic parameters. Methods: This study included healthy volunteers with full-body biplanar radiograph including body envelope reconstruction, allowing the estimation of COM location. The following parameters were analyzed: lumbar lordosis (LL), thoracic kyphosis (TK), cervical lordosis (CL), pelvic tilt (PT), Sacro-femoral angle (SFA), Knee flexion angle (KFA), sagittal odontoid-hip axis angle (ODHA). The following COM in the sagittal plane were located: whole body, at thoracolumbar inflexion point, and body segment above TK apex. The body envelope reconstruction also provided the distance between anterior skin and the LL apex vertebral body center (“SV-L distance”). Results: This study included 124 volunteers, with a mean age of 44±19.3. Multivariate analysis confirmed posterior translation of COM above TK apex with increasing LL (P=0.002) through its proximal component, and posterior shift of COM at inflexion point with increasing TK (P=0.008). Increased SV-L distance was associated with greater ODHA (r=0.4) and more anterior body COM (r=0.8), caused by increased TK (r=0.2) and decreased proximal and distal LL (both r=0.3), resulting in an augmentation in SFA (r=0.3) (all P<0.01). Conclusions: Barycentremetry showed that greater LL was associated with posterior shift of COM above thoracic apex while greater TK was correlated with more posterior COM at inflexion point. Whole-body COM was strongly correlated with ODHA. This study also exhibited significant alignment disruption associated with increased abdominal volume, with compensatory hip extension. Level of evidence: II Mon, 01 Apr 2024 00:00:00 GMT http://hdl.handle.net/10985/25103 2024-04-01T00:00:00Z KHALIFE, Marc VERGARI, Claudio ASSI, Ayman GUIGUI, Pierre ATTALI, Valerie VALENTIN, Rémi VAFADAR, Saman FERRERO, Emmanuelle SKALLI, Wafa Study design: Multicentric retrospective. Objective: The study of center of mass (COM) locations (i.e. barycentremetry) can help us understand postural alignment. This study goal was to determine relationships between COM locations and global postural alignment X-ray parameters in healthy subjects. The second objective was to determine the impact on spinopelvic alignment of increased distance between anterior body envelope and spine at lumbar apex level. Summary of background data: Unexplored relationship between COM location and spinopelvic parameters. Methods: This study included healthy volunteers with full-body biplanar radiograph including body envelope reconstruction, allowing the estimation of COM location. The following parameters were analyzed: lumbar lordosis (LL), thoracic kyphosis (TK), cervical lordosis (CL), pelvic tilt (PT), Sacro-femoral angle (SFA), Knee flexion angle (KFA), sagittal odontoid-hip axis angle (ODHA). The following COM in the sagittal plane were located: whole body, at thoracolumbar inflexion point, and body segment above TK apex. The body envelope reconstruction also provided the distance between anterior skin and the LL apex vertebral body center (“SV-L distance”). Results: This study included 124 volunteers, with a mean age of 44±19.3. Multivariate analysis confirmed posterior translation of COM above TK apex with increasing LL (P=0.002) through its proximal component, and posterior shift of COM at inflexion point with increasing TK (P=0.008). Increased SV-L distance was associated with greater ODHA (r=0.4) and more anterior body COM (r=0.8), caused by increased TK (r=0.2) and decreased proximal and distal LL (both r=0.3), resulting in an augmentation in SFA (r=0.3) (all P<0.01). Conclusions: Barycentremetry showed that greater LL was associated with posterior shift of COM above thoracic apex while greater TK was correlated with more posterior COM at inflexion point. Whole-body COM was strongly correlated with ODHA. This study also exhibited significant alignment disruption associated with increased abdominal volume, with compensatory hip extension. Level of evidence: II http://hdl.handle.net/10985/20156 A novel dataset and deep learning-based approach for marker-less motion capture during gait VAFADAR, Saman; SKALLI, Wafa; BONNET-LEBRUN, Aurore; KHALIFÉ, Marc; RENAUDIN, Mathis; HAMZA, Amine; GAJNY, Laurent Background: The deep learning-based human pose estimation methods, which can estimate joint centers position, have achieved promising results on the publicly available human pose datasets (e.g., Human3.6 M). However, these datasets may be less efficient for gait study, particularly for clinical applications, because of the limited number of subjects, their homogeneity (all asymptomatic adults), and the errors introduced by marker placement on subjects’ regular clothing. Research question: How a new human pose dataset, adapted for gait study, could contribute to the advancement and evaluation of marker-less motion capture systems? Methods: A marker-less system, based on deep learning-based pose estimation methods, was proposed. A new dataset (ENSAM dataset) was collected. Twenty-two asymptomatic adults, one adult with scoliosis, one adult with spondylolisthesis, and seven children with bone disease performed ten walking trials, while being recorded both by the proposed marker-less system and a reference system – combining a marker-based motion capture system and a medical imaging system (EOS). The dataset was split into training and test sets. The pose estimation method, already trained on the Human3.6 M dataset, was evaluated on the ENSAM test set, then reevaluated after further training on the ENSAM training set. The joints coordinates were evaluated, using Bland-Altman bias and 95 % confidence interval, and joint position error (the Euclidean distance between the estimated joint centers and the corresponding reference values). Results: The Bland-Altman 95 % confidence intervals were substantially improved after finetuning the pose estimation method on the ENSAM training set (e.g., from 106.9 mm to 17.4 mm for the hip joint). With the new dataset and approach, the mean joint position error varied from 6.2 mm for ankles to 21.1 mm for shoulders. Significance: The proposed marker-less system achieved promising results in terms of joint position errors. Future studies are necessary to assess the system in terms of gait parameters. Fri, 01 Jan 2021 00:00:00 GMT http://hdl.handle.net/10985/20156 2021-01-01T00:00:00Z VAFADAR, Saman SKALLI, Wafa BONNET-LEBRUN, Aurore KHALIFÉ, Marc RENAUDIN, Mathis HAMZA, Amine GAJNY, Laurent Background: The deep learning-based human pose estimation methods, which can estimate joint centers position, have achieved promising results on the publicly available human pose datasets (e.g., Human3.6 M). However, these datasets may be less efficient for gait study, particularly for clinical applications, because of the limited number of subjects, their homogeneity (all asymptomatic adults), and the errors introduced by marker placement on subjects’ regular clothing. Research question: How a new human pose dataset, adapted for gait study, could contribute to the advancement and evaluation of marker-less motion capture systems? Methods: A marker-less system, based on deep learning-based pose estimation methods, was proposed. A new dataset (ENSAM dataset) was collected. Twenty-two asymptomatic adults, one adult with scoliosis, one adult with spondylolisthesis, and seven children with bone disease performed ten walking trials, while being recorded both by the proposed marker-less system and a reference system – combining a marker-based motion capture system and a medical imaging system (EOS). The dataset was split into training and test sets. The pose estimation method, already trained on the Human3.6 M dataset, was evaluated on the ENSAM test set, then reevaluated after further training on the ENSAM training set. The joints coordinates were evaluated, using Bland-Altman bias and 95 % confidence interval, and joint position error (the Euclidean distance between the estimated joint centers and the corresponding reference values). Results: The Bland-Altman 95 % confidence intervals were substantially improved after finetuning the pose estimation method on the ENSAM training set (e.g., from 106.9 mm to 17.4 mm for the hip joint). With the new dataset and approach, the mean joint position error varied from 6.2 mm for ankles to 21.1 mm for shoulders. Significance: The proposed marker-less system achieved promising results in terms of joint position errors. Future studies are necessary to assess the system in terms of gait parameters.