pith. sign in

arxiv: 2606.25284 · v1 · pith:KBM7CD3Onew · submitted 2026-06-24 · 💻 cs.CV

Evaluation Protocols and Validation for Cameras in Indoor Healthcare Monitoring

Amirhossein Dadashzadeh , Jingjing Liu , Qianhui Men , Qiushuo Cheng , Kirsty Scott , Lisa Alcock , Ian Craddock , Majid Mirmehdi This is my paper

Pith reviewed 2026-06-25 21:41 UTC · model grok-4.3

classification 💻 cs.CV
keywords camera validationhealthcare monitoringpose estimationRGB-D camerasmetrological performancedepth bias3D reconstructionindoor monitoring
0
0 comments X

The pith

Metrological performance varies substantially across RGB-D cameras, with depth bias at 5 m ranging from 50 mm to over 1400 mm and directly driving 3D pose reconstruction errors from 104 mm to 365 mm.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops two validation protocols to characterize cameras for indoor healthcare monitoring: one measuring device metrology and the other measuring support for human pose estimation. Five cameras are tested under controlled changes in lighting, height, angle, and occlusion in representative indoor scenes. Depth accuracy differs markedly by device while 2D pose accuracy remains broadly comparable across cameras and estimators. The resulting 3D errors track the underlying depth quality, and environmental factors affect 3D results in a device- and estimator-specific manner. The protocols supply concrete data for selecting cameras that match the accuracy needs of clinical or home monitoring tasks.

Core claim

We present two technical validation protocols that evaluate the metrological performance of RGB and RGB-D cameras and their use in supporting human pose estimation. Systematic assessment of five cameras under controlled variations in lighting, camera height, viewing angle, and occlusion level shows depth bias at 5 m ranging from 50 mm to over 1400 mm depending on the device. For 2D pose estimation all cameras achieve comparable accuracy with mean mAP between approximately 78% and 90%, whereas 3D reconstruction error differs markedly with MPJPE ranging from 104 mm to 365 mm, closely reflecting underlying depth-sensing quality. Environmental factors have a camera- and estimator-dependent effec

What carries the argument

Two technical validation protocols that separately quantify metrological performance and pose-estimation support under controlled indoor variations in lighting, height, angle, and occlusion.

Load-bearing premise

The controlled variations in lighting, camera height, viewing angle, and occlusion within representative indoor scenarios adequately capture the practical deployment factors encountered in real-world healthcare settings.

What would settle it

A deployment study in actual patient homes or clinics that measures depth bias and MPJPE under uncontrolled conditions and finds performance spreads that do not match the ordering or magnitudes observed in the controlled tests.

Figures

Figures reproduced from arXiv: 2606.25284 by Amirhossein Dadashzadeh, Ian Craddock, Jingjing Liu, Kirsty Scott, Lisa Alcock, Majid Mirmehdi, Qianhui Men, Qiushuo Cheng.

Figure 1
Figure 1. Figure 1: Overview of the proposed two-stage validation framework. Protocol 1 evaluates metrological perfor￾mance across five cameras; Protocol 2 assesses pose estimation accuracy for the cameras, validated against a motion capture system under controlled, indoor deployment conditions. 2. Related work Camera-based sensing has become an increasingly popular modality for indoor healthcare mon￾itoring, driven by advanc… view at source ↗
Figure 2
Figure 2. Figure 2: Experimental platform setup for measuring depth error. ated under controlled variations of both lighting conditions and camera-to-object distance. Ambient illumination is monitored using a lux meter to ensure consistency across experiments. Under normal indoor lighting conditions (450600 lux), depth is recorded at distances ranging from 1 to 5m in 1m in￾crements. To assess the influence of reduced illumina… view at source ↗
Figure 3
Figure 3. Figure 3: FOV angle measurements. (a) Acrylic sheet for FOV. (b) FOV angle calculations. is noted as the representative device temperature at each time point. Three experimental factors are considered, scene dynamics, camera operation, and ambient room temperature. These are examined through two separate experimental configurations as described below. Scene dynamics and camera operation: Cameras are deployed in the … view at source ↗
Figure 4
Figure 4. Figure 4: Template of marker arrangement used for the motion capture system, shown in front (left) and back (right) views. Marker labels use the following abbreviations: FHD (forehead), BHD (back of head), SHO (shoul￾der), ELB (elbow), WRI (wrist), KNEE (knee), and ANK (ankle). The suffixes L and R denote left and right sides of the body, respectively, while LAT and MID indicate lateral and medial marker positions. … view at source ↗
Figure 5
Figure 5. Figure 5: Experimental platform setup and motion task used for evaluating human pose estimation performance. The participant started from the chair and performed a sequence of motion tasks including (a) sitting for 3 s, (b) sit-to-stand, (c) straight walking along the designated path, (d) in-place turning, and (e) walking with a curved turn before returning to the starting point. The layout also shows the spatial ar… view at source ↗
Figure 6
Figure 6. Figure 6: Experimental setup of cameras and Vicon motion capture system. within the feasible range, providing a balanced margin with respect to both the upper and very low visibility constraints. The resulting angle ranges and recommended tilt angles are summarised in [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Demonstration of the feasible camera tilt angle range measurement. 3.2.3. Signal processing and synchronisation Data post-processing includes labelling marker trajectories automatically to start with and then inspecting them to ensure consistency. Temporal gaps in the marker data caused by occlusions were then manually filled within the Vicon Nexus software to reconstruct the full-body motion model. The RG… view at source ↗
Figure 8
Figure 8. Figure 8: COCO style 2D human pose represen￾tation [69] [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Bias of the tested cameras at different camera-to-target distances under different lighting condition. 1 2 3 4 5 Distance (m) 0 200 400 600 800 1000 1200 Depth Precision (mm) Depth Precision vs Distance (Normal Lighting) Camera Femto Bolt OAK-D Pro RealSense ZED 2 1 2 3 4 5 Distance (m) 0 200 400 600 800 1000 1200 Depth Precision (mm) Depth Precision vs Distance (Low Lighting) Camera Femto Bolt OAK-D Pro R… view at source ↗
Figure 10
Figure 10. Figure 10: Precision of the tested cameras at different camera-to-target distances under different lighting condi￾tion. 4.1. Metrological Performance Evaluation Depth measurement performance [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Temperatures of tested cameras after running for 2h in different room temperatures. Femto Bolt Oak-D pro Real Sense ZED2 Camera 0 10 20 30 40 50 60 RMSE (mm) RMSE by Camera and Temperature Temperature Low Normal High Femto Bolt Oak-D pro Real Sense ZED2 Camera 0 5 10 15 20 25 30 35 40 Drift (mm) Average Absolute Drift by Camera and Temperature Temperature Low Normal High [PITH_FULL_IMAGE:figures/full_fig… view at source ↗
Figure 12
Figure 12. Figure 12: The stability of tested cameras after running for 2h in different room temperatures. precision values that lie between those of the Femto Bolt and OAK-D Pro. Under very low lighting conditions or at short distances, however, the precision of OAK-D Pro becomes comparable to that of RealSense and ZED 2. Temperature and stability performance – Temperature is evaluated for the four RGB-D cameras un￾der two in… view at source ↗
Figure 13
Figure 13. Figure 13: Comparison of measured and manufacturer-specified DFOV across multiple camera models and resolutions. Percentage annotations show how each measurement compares to the corresponding manufacturer reported. D: Diagonal, W: Widest, N: Narrowest. When comparing all cameras at the standard resolution of 1280×720, the field of view ranking from widest to narrowest is given in [PITH_FULL_IMAGE:figures/full_fig_p… view at source ↗
Figure 14
Figure 14. Figure 14: Comparative evaluation of the four depth cameras across key performance metrics. For depth bias, temperature rise, and runtime RMSE, very low values indicate better performance. In contrast, higher values are preferable for measured FOV [PITH_FULL_IMAGE:figures/full_fig_p016_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Overview of a qualitative score comparison for the four depth cameras. (1.8m, 2.0m, and 2.2m), selected to represent practical wall-mounted installation positions in typical residential environments where standard ceiling heights are approximately 2.4m. As demonstrated by the field-of-view analysis in Section 3.2.2 and [PITH_FULL_IMAGE:figures/full_fig_p017_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: 2D pose estimation performance across the four cameras for two pose estimators: (a) RTMO and (b) YOLO26. In each row, the left and centre panels show the impact of lighting and furniture occlusion on mAP@(0.5,0.95) and PCK@0.2 respectively (averaged over mounting height), and the right panel shows the im￾pact of camera mounting height on mAP (averaged over furniture). occlusion-related reductions are obse… view at source ↗
Figure 17
Figure 17. Figure 17: 3D pose estimation performance across the three RGBD cameras for two pose estimators: (a) RTMO and (b) YOLO26. In each row, the left and centre panels show the impact of lighting and furniture occlusion on MPJPE and PCK respectively (averaged over mounting height), and the right panel shows the impact of camera mounting height on MPJPE (averaged over furniture). Lower MPJPE is better. The results demonstr… view at source ↗
read the original abstract

Camera-based monitoring systems are increasingly adopted in healthcare settings for the continuous assessment of patient movement and activities. However, their technical performance under real-world indoor conditions remains insufficiently characterised, preventing appropriate camera selection for clinical or home adoption and reproducibility. Existing validation studies typically assess either device metrological performance or algorithm accuracy in isolation, and often do not systematically account for practical deployment factors, such as lighting variability, occlusions, and camera positioning. We present two technical validation protocols: the first evaluates the metrological performance of RGB and RGB-D cameras, and the second assesses their use in supporting human pose estimation, validated using state-of-the-art pose estimators. The proposed protocols systematically assess five cameras, four RGB-D and one RGB, under controlled variations in lighting, camera height, viewing angle, and occlusion level within representative indoor scenarios. The experimental results show that metrological performance varies substantially across cameras, with depth bias at 5 m ranging from 50 mm to over 1400 mm depending on the device. For 2D pose estimation, all cameras achieve broadly comparable accuracy, with mean mAP between approximately 78% and 90% across cameras and estimators, whereas 3D reconstruction error differs markedly across devices, with MPJPE ranging from 104 mm to 365 mm, closely reflecting underlying depth-sensing quality. Environmental factors have a camera- and estimator-dependent effect on 3D performance, while camera mounting height has minimal influence within the evaluated range. This work provides evidence-based guidance for the selection and deployment of cameras in healthcare monitoring applications, addressing an important gap in current technical validation practice.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper presents two technical validation protocols for RGB and RGB-D cameras in indoor healthcare monitoring: one assessing metrological performance (e.g., depth bias) and the second evaluating support for human pose estimation using state-of-the-art estimators. Five cameras are tested under controlled variations in lighting, camera height, viewing angle, and occlusion within representative indoor scenarios. Key results include substantial cross-device variation in depth bias at 5 m (50 mm to >1400 mm), comparable 2D mAP (78–90%), and markedly different 3D MPJPE (104–365 mm) that tracks depth quality; environmental factors show camera- and estimator-dependent effects while mounting height has minimal influence. The work claims to provide evidence-based guidance for camera selection and deployment, addressing gaps in existing validation practice.

Significance. If the protocols adequately represent practical conditions, the work is significant for filling a gap in systematic, reproducible validation that jointly considers device metrology and algorithm performance under deployment-relevant factors. The explicit linkage between depth-sensing quality and 3D reconstruction error, plus the finding of minimal height influence, offers actionable insights for healthcare applications. The provision of two distinct protocols is a strength for future reproducibility.

major comments (1)
  1. [Abstract / Methods (experimental design)] Abstract / Methods (experimental design paragraph): The central claim that the protocols supply 'evidence-based guidance for the selection and deployment of cameras in healthcare monitoring applications' rests on the tested variations (lighting, height, angle, occlusion) in 'representative indoor scenarios' being sufficient. However, untested factors such as dynamic multi-person scenes, IR interference from medical devices, long-term thermal drift, and cluttered multi-occupant rooms are not addressed; if these dominate real deployments, the reported ranges (depth bias 50–1400 mm, MPJPE 104–365 mm) and the assertion that 3D error 'closely reflects' depth quality lose generalizability.
minor comments (1)
  1. [Abstract] Abstract: ranges for mAP ('approximately 78% to 90%') and MPJPE are stated without per-camera values, error bars, or statistical tests; adding a summary table with exact figures and variability measures would strengthen the quantitative claims.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the scope of our experimental design and the generalizability of our claims. We respond point-by-point below.

read point-by-point responses

  1. Referee: [Abstract / Methods (experimental design)] Abstract / Methods (experimental design paragraph): The central claim that the protocols supply 'evidence-based guidance for the selection and deployment of cameras in healthcare monitoring applications' rests on the tested variations (lighting, height, angle, occlusion) in 'representative indoor scenarios' being sufficient. However, untested factors such as dynamic multi-person scenes, IR interference from medical devices, long-term thermal drift, and cluttered multi-occupant rooms are not addressed; if these dominate real deployments, the reported ranges (depth bias 50–1400 mm, MPJPE 104–365 mm) and the assertion that 3D error 'closely reflects' depth quality lose generalizability.

    Authors: We agree that the tested factors do not exhaustively represent all conditions that may occur in healthcare deployments, and therefore the specific numerical ranges and the strength of the 'closely reflects' linkage are scoped to our controlled single-person indoor setups. The variations we chose (lighting, height, angle, occlusion) were selected because they are common, measurable, and directly affect both metrological accuracy and pose estimation; they also allow reproducible protocol definition. Factors such as dynamic multi-person scenes, IR interference from medical devices, thermal drift over time, and cluttered multi-occupant rooms would require distinct experimental extensions (e.g., multi-person datasets or long-duration thermal cycling) that fall outside the current protocol scope. We will make a partial revision: (1) add an explicit Limitations subsection in the Discussion that lists these untested factors and states that the reported ranges apply to the evaluated conditions; (2) revise the abstract claim from 'provides evidence-based guidance' to 'provides initial evidence-based guidance for the tested deployment factors'. This clarifies the contribution without overstating generalizability while retaining the value of the two protocols and the observed depth-to-3D-error relationship within the tested regime. revision: partial

Circularity Check

0 steps flagged

No circularity: purely empirical measurements with no derivations or self-referential claims

full rationale

The paper presents two validation protocols and reports direct experimental measurements of camera metrological performance and pose estimation accuracy under controlled conditions. No equations, fitted parameters, predictions, or mathematical derivations are present. All reported values (depth bias ranges, mAP, MPJPE) are stated as observed results from testing five cameras. No self-citations are invoked as load-bearing for any claim, and the central results do not reduce to inputs by construction. This matches the default expectation for an empirical comparison paper.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work rests on standard computer-vision domain assumptions with no free parameters, invented entities, or ad-hoc axioms introduced to support the central claims.

axioms (1)
  • domain assumption State-of-the-art pose estimators provide a valid proxy for assessing camera utility in 3D reconstruction tasks.
    The second protocol relies on existing pose estimators to evaluate camera performance without independent verification of estimator accuracy in the tested conditions.

pith-pipeline@v0.9.1-grok · 5846 in / 1279 out tokens · 33830 ms · 2026-06-25T21:41:37.528297+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

76 extracted references · 3 canonical work pages

  1. [1]

    Gait speed and survival in older adults

    Stephanie Studenski, Subashan Perera, Kushang Patel, Caterina Rosano, Kimberly Faulkner, Marco Inzitari, Jennifer Brach, Julie Chandler, Peggy Cawthon, Elizabeth Barrett Connor, et al. Gait speed and survival in older adults. Jama, 305(1):50–58, 2011

    2011

  2. [2]

    Moving forward on gait measurement: toward a more refined approach

    Sue Lord, Brook Galna, and Lynn Rochester. Moving forward on gait measurement: toward a more refined approach. Movement Disorders, 28(11):1534–1543, 2013

    2013

  3. [3]

    Sit-to-stand analysis in the wild using silhouettes for longitudinal health monitoring

    Alessandro Masullo, Tilo Burghardt, Toby Perrett, Dima Damen, and Majid Mirmehdi. Sit-to-stand analysis in the wild using silhouettes for longitudinal health monitoring. In International Conference on Image Analysis and Recognition , pages 175–185. Springer, 2019

    2019

  4. [4]

    Protocol for pd sensors: Parkinsons disease symptom 25 of 28 evaluation in a naturalistic setting producing outcome measures using sphere technology

    Catherine Morgan, Ian Craddock, Emma L Tonkin, Kirsi M Kinnunen, Roisin McNaney , Sam Whitehouse, Majid Mirme- hdi, Farnoosh Heidarivincheh, Ryan McConville, Julia Carey , et al. Protocol for pd sensors: Parkinsons disease symptom 25 of 28 evaluation in a naturalistic setting producing outcome measures using sphere technology . an observational feasibilit...

    2020

  5. [5]

    Bridging e-health and the internet of things: The sphere project

    Ni Zhu, Tom Diethe, Massimo Camplani, Lili Tao, Alison Burrows, Niall Twomey , Dritan Kaleshi, Majid Mirmehdi, Peter Flach, and Ian Craddock. Bridging e-health and the internet of things: The sphere project. IEEE Intelligent Systems, 30(4):39–46, 2015. doi: 10.1109/MIS.2015.57

    work page doi:10.1109/mis.2015.57 2015

  6. [6]

    Multimodal indoor localisation in parkinson’s disease for detecting medication use: Observational pilot study in a free-living setting

    Ferdian Jovan, Catherine Morgan, Ryan McConville, Emma L Tonkin, Ian Craddock, and Alan Whone. Multimodal indoor localisation in parkinson’s disease for detecting medication use: Observational pilot study in a free-living setting. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages 4273–4283, 2023

    2023

  7. [7]

    Your turn: At home turning angle estimation for parkinsons disease severity assessment

    Qiushuo Cheng, Catherine Morgan, Arindam Sikdar, Alessandro Masullo, Alan Whone, and Majid Mirmehdi. Your turn: At home turning angle estimation for parkinsons disease severity assessment. Artificial Intelligence in Medicine, 167: 103194, 2025. ISSN 0933-3657. doi: https://doi.org/10.1016/j.artmed.2025.103194

    work page doi:10.1016/j.artmed.2025.103194 2025

  8. [8]

    Real-world wrist-derived digital mobility outcomes in people with multiple long-term conditions: A comparison of algorithms

    Dimitrios Megaritis, Lisa Alcock, Kirsty Scott, Hugo Hiden, Andrea Cereatti, Ioannis V ogiatzis, and Silvia Del Din. Real-world wrist-derived digital mobility outcomes in people with multiple long-term conditions: A comparison of algorithms. Bioengineering, 12(10):1108, 2025

    2025

  9. [9]

    Advancements in wearable sensor technologies for health monitoring in terms of clinical applications, rehabilitation, and disease risk assessment: Systematic review

    Bonsang Gu, Hyeon Su Kim, HyunBin Kim, and Jun-Il Yoo. Advancements in wearable sensor technologies for health monitoring in terms of clinical applications, rehabilitation, and disease risk assessment: Systematic review. JMIR mHealth and uHealth, 14(1):e76084, 2026

    2026

  10. [10]

    Advancing real-world applications: A scoping review on emerging wearable technologies for recognizing activities of daily living

    Mustafa Elhadi Ahmed, Hongnian Yu, Michael Vassallo, and Pelagia Koufaki. Advancing real-world applications: A scoping review on emerging wearable technologies for recognizing activities of daily living. Smart Health, page 100555, 2025

    2025

  11. [11]

    Technical validation of real-world monitoring of gait: a multicentric observational study

    Claudia Mazzà, Lisa Alcock, Kamiar Aminian, Clemens Becker, Stefano Bertuletti, Tecla Bonci, Philip Brown, Marina Brozgol, Ellen Buckley , Anne-Elie Carsin, et al. Technical validation of real-world monitoring of gait: a multicentric observational study . BMJ open, 11(12):e050785, 2021

    2021

  12. [12]

    A review of the evolution of vision-based motion analysis and the integration of advanced computer vision methods towards developing a markerless system

    Steffi L Colyer, Murray Evans, Darren P Cosker, and Aki IT Salo. A review of the evolution of vision-based motion analysis and the integration of advanced computer vision methods towards developing a markerless system. Sports medicine-open, 4(1):24, 2018

    2018

  13. [13]

    Rgb-d data-based action recognition: a review

    Muhammad Bilal Shaikh and Douglas Chai. Rgb-d data-based action recognition: a review. Sensors, 21(12):4246, 2021

    2021

  14. [14]

    Multimodal classification of parkinsons disease in home environments with resiliency to missing modalities

    Farnoosh Heidarivincheh, Ryan McConville, Catherine Morgan, Roisin McNaney , Alessandro Masullo, Majid Mirme- hdi, Alan L Whone, and Ian Craddock. Multimodal classification of parkinsons disease in home environments with resiliency to missing modalities. Sensors, 21(12):4133, 2021

    2021

  15. [15]

    Automated real-world video analysis of sit-to-stand transitions predicts parkinsons disease severity

    Catherine Morgan, Alessandro Masullo, Majid Mirmehdi, Hanna Kristiina Isotalus, Ferdian Jovan, Ryan McConville, Emma L Tonkin, Alan Whone, and Ian Craddock. Automated real-world video analysis of sit-to-stand transitions predicts parkinsons disease severity . Digital Biomarkers, 7(1):92–103, 2023

    2023

  16. [16]

    Person re-id by fusion of video silhouettes and wearable signals for home monitoring applications

    Alessandro Masullo, Tilo Burghardt, Dima Damen, Toby Perrett, and Majid Mirmehdi. Person re-id by fusion of video silhouettes and wearable signals for home monitoring applications. Sensors, 20(9):2576, 2020

    2020

  17. [17]

    No need for a lab: Towards multi-sensory fusion for ambient assisted living in real-world living homes

    Alessandro Masullo, Toby Perrett, Dima Damen, Tilo Burghardt, and Majid Mirmehdi. No need for a lab: Towards multi-sensory fusion for ambient assisted living in real-world living homes. In VISIGRAPP (5: VISAPP) , pages 328–337, 2021

    2021

  18. [18]

    A metrological and application-related comparison of six consumer grade stereo depth cameras for the use in robotics

    Michel Heinemann, Jonas Herzfeld, Martin Sliwinski, Johannes Hinckeldeyn, and Jochen Kreutzfeldt. A metrological and application-related comparison of six consumer grade stereo depth cameras for the use in robotics. In 2022 IEEE International Symposium on Robotic and Sensors Environments (ROSE) , pages 01–07. IEEE, 2022

    2022

  19. [19]

    Performance evaluation of the 1st and 2nd generation kinect for multimedia applications

    Simone Zennaro, Matteo Munaro, Simone Milani, Pietro Zanuttigh, Andrea Bernardi, Stefano Ghidoni, and Emanuele Menegatti. Performance evaluation of the 1st and 2nd generation kinect for multimedia applications. In 2015 IEEE International Conference on Multimedia and Expo (ICME) , pages 1–6. IEEE, 2015

    2015

  20. [20]

    Metro- logical characterization and comparison of d415, d455, l515 realsense devices in the close range

    Michaela Servi, Elisa Mussi, Andrea Profili, Rocco Furferi, Yary V olpe, Lapo Governi, and Francesco Buonamici. Metro- logical characterization and comparison of d415, d455, l515 realsense devices in the close range. Sensors, 21(22):7770, 2021

    2021

  21. [21]

    Video- based pose estimation for gait analysis in stroke survivors during clinical assessments: a proof-of-concept study

    Luca Lonini, Yaejin Moon, Kyle Embry , R James Cotton, Kelly McKenzie, Sophia Jenz, and Arun Jayaraman. Video- based pose estimation for gait analysis in stroke survivors during clinical assessments: a proof-of-concept study . Digital Biomarkers, 6(1):9–18, 2022. 26 of 28

    2022

  22. [22]

    Influence of the camera viewing angle on openpose validity in motion analysis

    Melanie Baldinger, Lara Marie Reimer, and V eit Senner. Influence of the camera viewing angle on openpose validity in motion analysis. Sensors (Basel, Switzerland), 25(3):799, 2025

    2025

  23. [23]

    Design and validation of depth camera-based static posture assessment system

    Qingjun Xing, Ruiwei Hong, Yuanyuan Shen, and Yanfei Shen. Design and validation of depth camera-based static posture assessment system. Iscience, 26(10), 2023

    2023

  24. [24]

    Validation of a single rgb-d camera for gait assessment of polyneuropathy patients

    Maria do Carmo Vilas-Boas, Ana Patrícia Rocha, Hugo Miguel Pereira Choupina, Márcio Neves Cardoso, José Maria Fernandes, Teresa Coelho, and João Paulo Silva Cunha. Validation of a single rgb-d camera for gait assessment of polyneuropathy patients. Sensors, 19(22):4929, 2019

    2019

  25. [25]

    Validation of a 3d markerless system for gait analysis based on openpose and two rgb webcams

    Erika DAntonio, Juri Taborri, Ilaria Mileti, Stefano Rossi, and Fabrizio Patané. Validation of a 3d markerless system for gait analysis based on openpose and two rgb webcams. IEEE Sensors Journal, 21(15):17064–17075, 2021

    2021

  26. [26]

    Comparative abilities of microsoft kinect and vicon 3d motion capture for gait analysis

    Alexandra Pfister, Alexandre M West, Shaw Bronner, and Jack Adam Noah. Comparative abilities of microsoft kinect and vicon 3d motion capture for gait analysis. Journal of medical engineering & technology , 38(5):274–280, 2014

    2014

  27. [27]

    Sofia Scataglini, Eveline Abts, Cas Van Bocxlaer, Maxime Van den Bussche, Sara Meletani, and Steven Truijen. Accuracy , validity , and reliability of markerless camera-based 3d motion capture systems versus marker-based 3d motion capture systems in gait analysis: a systematic review and meta-analysis. Sensors, 24(11):3686, 2024

    2024

  28. [28]

    Rtmo: Towards high-performance one-stage real-time multi-person pose estimation

    Peng Lu, Tao Jiang, Yining Li, Xiangtai Li, Kai Chen, and Wenming Yang. Rtmo: Towards high-performance one-stage real-time multi-person pose estimation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1491–1500, 2024

    2024

  29. [29]

    Yolo26: key architectural enhancements and performance benchmarking for real-time object detection

    Ranjan Sapkota, Rahul Harsha Cheppally , Ajay Sharda, and Manoj Karkee. Yolo26: key architectural enhancements and performance benchmarking for real-time object detection. arXiv preprint arXiv:2509.25164, 2025

    arXiv 2025

  30. [30]

    A computer vision system for deep learning-based detection of patient mobilization activities in the icu

    Serena Yeung, Francesca Rinaldo, Jeffrey Jopling, Bingbin Liu, Rishab Mehra, N Lance Downing, Michelle Guo, Gabriel M Bianconi, Alexandre Alahi, Julia Lee, et al. A computer vision system for deep learning-based detection of patient mobilization activities in the icu. NPJ digital medicine, 2(1):11, 2019

    2019

  31. [31]

    Vision-based human activity recognition system using depth silhou- ettes: A smart home system for monitoring the residents

    Kibum Kim, Ahmad Jalal, and Maria Mahmood. Vision-based human activity recognition system using depth silhou- ettes: A smart home system for monitoring the residents. Journal of Electrical Engineering & T echnology, 14(6):2567–2573, 2019

    2019

  32. [32]

    Contactless patient care using hospital iot: Cctv-camera-based physiological monitoring in icu

    Haowen Wang, Jia Huang, Guowei Wang, Hongzhou Lu, and Wenjin Wang. Contactless patient care using hospital iot: Cctv-camera-based physiological monitoring in icu. IEEE Internet of Things Journal , 11(4):5781–5797, 2023

    2023

  33. [33]

    Generalized camera-based infant sleep-wake monitoring in nicus: A multi-center clinical trial

    Dongmin Huang, Dongfang Yu, Yongshen Zeng, Xiaoyan Song, Liping Pan, Junli He, Lirong Ren, Jie Yang, Hongzhou Lu, and Wenjin Wang. Generalized camera-based infant sleep-wake monitoring in nicus: A multi-center clinical trial. IEEE Journal of Biomedical and Health Informatics , 28(5):3015–3028, 2024

    2024

  34. [34]

    Smarthome-bench: A comprehensive benchmark for video anomaly detection in smart homes using multi-modal large language models

    Xinyi Zhao, Congjing Zhang, Pei Guo, Wei Li, Lin Chen, Chaoyue Zhao, and Shuai Huang. Smarthome-bench: A comprehensive benchmark for video anomaly detection in smart homes using multi-modal large language models. In Proceedings of the Computer Vision and Pattern Recognition Conference , pages 3975–3985, 2025

    2025

  35. [35]

    Empowering home security through wall crossing activity detection using vision cameras and convolutional long short-term architecture

    Muhammad Omair Khan, Haleem Farman, Md Ariful Islam Mozumder, Bilal Jan, Moustafa M Nasralla, and Hee-Cheol Kim. Empowering home security through wall crossing activity detection using vision cameras and convolutional long short-term architecture. IEEE Sensors Journal, 2025

    2025

  36. [36]

    Inter-trial variability is higher in 3d markerless compared to marker-based motion capture: Implications for data post-processing and analysis

    Brian Horsak, Kerstin Prock, Philipp Krondorfer, Tarique Siragy , Mark Simonlehner, and Bernhard Dumphart. Inter-trial variability is higher in 3d markerless compared to marker-based motion capture: Implications for data post-processing and analysis. Journal of Biomechanics, 166:112049, 2024

    2024

  37. [37]

    A systematic review of the applications of markerless motion capture (mmc) technology for clinical measurement in rehabilitation

    Winnie WT Lam, Yuk Ming Tang, and Kenneth NK Fong. A systematic review of the applications of markerless motion capture (mmc) technology for clinical measurement in rehabilitation. Journal of NeuroEngineering and Rehabilitation , 20 (1):57, 2023

    2023

  38. [38]

    A multimodal dataset of real-world mobility activities in Parkinsons disease

    Catherine Morgan, Emma L Tonkin, Alessandro Masullo, Ferdian Jovan, Arindam Sikdar, Pushpajit Khaire, Majid Mirmehdi, Ryan McConville, Gregory JL Tourte, Alan Whone, and Ian Craddock. A multimodal dataset of real-world mobility activities in Parkinsons disease. Scientific Data, 2023

    2023

  39. [39]

    Impersonal: an iot- aided computer vision framework for social distancing for health safety

    Romeo Giuliano, Eros Innocenti, Franco Mazzenga, Anna Maria V egni, and Alessandro Vizzarri. Impersonal: an iot- aided computer vision framework for social distancing for health safety . IEEE Internet of Things Journal , 9(10):7261–7272, 2021

    2021

  40. [40]

    A webcam-based machine learning approach for three-dimensional range of motion evaluation

    Xiaoye Michael Wang, Derek T Smith, and Qin Zhu. A webcam-based machine learning approach for three-dimensional range of motion evaluation. PLoS one, 18(10):e0293178, 2023. 27 of 28

    2023

  41. [41]

    Pierce, Daniel K

    Claire Chambers, Nidhi Seethapathi, Rachit Saluja, Helen Loeb, Samuel R. Pierce, Daniel K. Bogen, Laura Prosser, Michelle J. Johnson, and Konrad P . Kording. Computer vision to automatically assess infant neuromotor risk. IEEE T ransactions on Neural Systems and Rehabilitation Engineering, 28(11):2431–2442, 2020. doi: 10.1109/TNSRE.2020.3029121

    work page doi:10.1109/tnsre.2020.3029121 2020

  42. [42]

    Supervised classification of bradykinesia in parkinsons disease from smartphone videos

    Stefan Williams, Samuel D Relton, Hui Fang, Jane Alty , Rami Qahwaji, Christopher D Graham, and David C Wong. Supervised classification of bradykinesia in parkinsons disease from smartphone videos. Artificial Intelligence in Medicine, 110:101966, 2020

    2020

  43. [43]

    Bradykinesia recognition in parkinsons disease via single rgb video

    Bo Lin, Wei Luo, Zhiling Luo, Bo Wang, Shuiguang Deng, Jianwei Yin, and Mengchu Zhou. Bradykinesia recognition in parkinsons disease via single rgb video. ACM T ransactions on Knowledge Discovery from Data (TKDD), 14(2):1–19, 2020

    2020

  44. [44]

    Development of a low-cost markerless optical motion capture system for gait analysis and anthropometric parameter quantification

    Laura Alejandra Espitia-Mora, Manuel Andrés Vélez-Guerrero, and Mauro Callejas-Cuervo. Development of a low-cost markerless optical motion capture system for gait analysis and anthropometric parameter quantification. Sensors, 24(11): 3371, 2024

    2024

  45. [45]

    Validation of a markerless motion capture system for centre of mass kinematic analysis

    Ruben Valenzuela, Javier Corral, Mikel Diez, Thomas Provot, Francisco J Campa, Saioa Herrero, Erik Macho, and Charles Pinto. Validation of a markerless motion capture system for centre of mass kinematic analysis. Biocybernet- ics and Biomedical Engineering , 45(2):278–286, 2025

    2025

  46. [46]

    Automating the timed up and go test using a depth camera

    Amandine Dubois, Titus Bihl, and Jean-Pierre Bresciani. Automating the timed up and go test using a depth camera. Sensors, 18(1):14, 2017

    2017

  47. [47]

    Two-dimensional video-based analysis of human gait using pose estimation

    Jan Stenum, Cristina Rossi, and Ryan T Roemmich. Two-dimensional video-based analysis of human gait using pose estimation. PLoS computational biology, 17(4):e1008935, 2021

    2021

  48. [48]

    Kinect-based assessment of lower limbs during gait in post-stroke hemiplegic patients: A narrative review

    Serena Cerfoglio, Claudia Ferraris, Luca Vismara, Gianluca Amprimo, Lorenzo Priano, Giuseppe Pettiti, Manuela Galli, Alessandro Mauro, and V eronica Cimolin. Kinect-based assessment of lower limbs during gait in post-stroke hemiplegic patients: A narrative review. Sensors, 22(13):4910, 2022

    2022

  49. [49]

    Estimation of gait parameters in healthy and hemiplegic individuals using azure kinect: a comparative study with the optoelectronic system

    Serena Cerfoglio, Claudia Ferraris, Luca Vismara, Gianluca Amprimo, Lorenzo Priano, Matteo Bigoni, Manuela Galli, Alessandro Mauro, and V eronica Cimolin. Estimation of gait parameters in healthy and hemiplegic individuals using azure kinect: a comparative study with the optoelectronic system. Frontiers in bioengineering and biotechnology, 12:1449680, 2024

    2024

  50. [50]

    Openpose: Realtime multi-person 2d pose estimation using part affinity fields

    Zhe Cao, Gines Hidalgo, Tomas Simon, Shih-En Wei, and Yaser Sheikh. Openpose: Realtime multi-person 2d pose estimation using part affinity fields. IEEE transactions on pattern analysis and machine intelligence , 43(1):172–186, 2019

    2019

  51. [51]

    Metrological and critical characterization of the intel d415 stereo depth camera

    Monica Carfagni, Rocco Furferi, Lapo Governi, Chiara Santarelli, Michaela Servi, Francesca Uccheddu, and Yary V olpe. Metrological and critical characterization of the intel d415 stereo depth camera. Sensors, 19(3):489, 2019

    2019

  52. [52]

    Ex- perimental procedure for the metrological characterization of time-of-flight cameras for human body 3d measurements

    Simone Pasinetti, Cristina Nuzzi, Alessandro Luchetti, Matteo Zanetti, Matteo Lancini, and Mariolino De Cecco. Ex- perimental procedure for the metrological characterization of time-of-flight cameras for human body 3d measurements. Sensors, 23(1):538, 2023

    2023

  53. [53]

    Depth accuracy analysis of the zed 2i stereo camera in an indoor environment

    Ahmed Abdelsalam, Mostafa Mansour, Jari Porras, and Ari Happonen. Depth accuracy analysis of the zed 2i stereo camera in an indoor environment. Robotics and Autonomous Systems , 179:104753, 2024

    2024

  54. [54]

    Eval- uation of the pose tracking performance of the azure kinect and kinect v2 for gait analysis in comparison with a gold standard: A pilot study

    Justin Amadeus Albert, Victor Owolabi, Arnd Gebel, Clemens Markus Brahms, Urs Granacher, and Bert Arnrich. Eval- uation of the pose tracking performance of the azure kinect and kinect v2 for gait analysis in comparison with a gold standard: A pilot study . Sensors, 20(18):5104, 2020

    2020

  55. [55]

    Sina Mehdizadeh, Hoda Nabavi, Andrea Sabo, Twinkle Arora, Andrea Iaboni, and Babak Taati. Concurrent validity of human pose tracking in video for measuring gait parameters in older adults: a preliminary analysis with multiple trackers, viewing angles, and walking directions. Journal of neuroengineering and rehabilitation , 18(1):139, 2021

    2021

  56. [56]

    Low light performance of digital still cameras

    Dietmar Wueller. Low light performance of digital still cameras. In Multimedia Content and Mobile Devices , volume 8667, pages 434–442. SPIE, 2013

    2013

  57. [57]

    Mobile phone camera benchmarking in low light environment

    V eli-Tapani Peltoketo. Mobile phone camera benchmarking in low light environment. In Image Quality and System Performance XII, volume 9396, pages 68–77. SPIE, 2015

    2015

  58. [58]

    How good are rgb cameras retrieving colors of natural scenes and paintings?a study based on hyperspectral imaging

    João MM Linhares, José AR Monteiro, Ana Bailão, Liliana Cardeira, Taisei Kondo, Shigeki Nakauchi, Marcello Picollo, Costanza Cucci, Andrea Casini, Lorenzo Stefani, et al. How good are rgb cameras retrieving colors of natural scenes and paintings?a study based on hyperspectral imaging. Sensors, 20(21):6242, 2020

    2020

  59. [59]

    Low light level digital camera quality assessment

    Branko Livada, Dragana Peri´ c, and Saša Vuji´ c. Low light level digital camera quality assessment. Scientific T echnical Review, 71(2):3–13, 2021

    2021

  60. [60]

    Experimental validation of depth cameras for the parameterization of functional balance of patients in clinical tests

    Francisco-Ángel Moreno, José Antonio Merchán-Baeza, Manuel González-Sánchez, Javier González-Jiménez, and Anto- nio I Cuesta-Vargas. Experimental validation of depth cameras for the parameterization of functional balance of patients in clinical tests. Sensors, 17(2):424, 2017. 28 of 28

    2017

  61. [61]

    Validation, reliability , and responsiveness outcomes of kinematic assessment with an rgb-d camera to analyze movement in subacute and chronic low back pain

    Manuel Trinidad-Fernández, David Beckwée, Antonio Cuesta-Vargas, Manuel González-Sánchez, Francisco-Angel Moreno, Javier González-Jiménez, Erika Joos, and Peter Vaes. Validation, reliability , and responsiveness outcomes of kinematic assessment with an rgb-d camera to analyze movement in subacute and chronic low back pain. Sensors, 20 (3):689, 2020

    2020

  62. [62]

    Full-body motion assessment: Concurrent validation of two body tracking depth sensors versus a gold standard system during gait

    Maria do Carmo Vilas-Boas, Hugo Miguel Pereira Choupina, Ana Patrícia Rocha, José Maria Fernandes, and João Paulo Silva Cunha. Full-body motion assessment: Concurrent validation of two body tracking depth sensors versus a gold standard system during gait. Journal of biomechanics, 87:189–196, 2019

    2019

  63. [63]

    Validation of an ambient system for the measurement of gait parameters

    Amandine Dubois and Jean-Pierre Bresciani. Validation of an ambient system for the measurement of gait parameters. Journal of biomechanics, 69:175–180, 2018

    2018

  64. [64]

    Deeplabcut: markerless pose estimation of user-defined body parts with deep learning

    Alexander Mathis, Pranav Mamidanna, Kevin M Cury , Taiga Abe, V enkatesh N Murthy , Mackenzie Weygandt Mathis, and Matthias Bethge. Deeplabcut: markerless pose estimation of user-defined body parts with deep learning. Nature neuroscience, 21(9):1281–1289, 2018

    2018

  65. [65]

    Validity of ai-based gait analysis for simultaneous measurement of bilateral lower limb kinematics using a single video camera

    Takumi Ino, Mina Samukawa, Tomoya Ishida, Naofumi Wada, Yuta Koshino, Satoshi Kasahara, and Harukazu Tohyama. Validity of ai-based gait analysis for simultaneous measurement of bilateral lower limb kinematics using a single video camera. Sensors, 23(24):9799, 2023

    2023

  66. [66]

    Vision-based method for automatic quantification of parkinsonian bradykinesia

    Yu Liu, Jiansheng Chen, Chunhua Hu, Yu Ma, Dongyun Ge, Suhua Miao, Youze Xue, and Luming Li. Vision-based method for automatic quantification of parkinsonian bradykinesia. IEEE T ransactions on Neural Systems and Rehabilitation Engineering, 27(10):1952–1961, 2019

    1952

  67. [67]

    Robust self-adaptation fall-detection system based on camera height

    Xiangbo Kong, Lehan Chen, Zhichen Wang, Yuxi Chen, Lin Meng, and Hiroyuki Tomiyama. Robust self-adaptation fall-detection system based on camera height. Sensors, 19(17):3768, 2019

    2019

  68. [68]

    A method to compensate for the errors caused by temperature in structured-light 3d cameras

    Oriol Vila, Imma Boada, David Raba, and Esteve Farres. A method to compensate for the errors caused by temperature in structured-light 3d cameras. Sensors, 21(6):2073, 2021

    2073

  69. [69]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In European conference on computer vision , pages 740–755. Springer, 2014

    2014

  70. [70]

    A comparison of the accuracy of several hip center location prediction methods

    Alexander L Bell, Douglas R Pedersen, and Richard A Brand. A comparison of the accuracy of several hip center location prediction methods. Journal of biomechanics, 23(6):617–621, 1990

    1990

  71. [71]

    Mediapipe: A framework for building perception pipelines

    Camillo Lugaresi, Jiuqiang Tang, Hadon Nash, Chris McClanahan, Esha Uboweja, Michael Hays, Fan Zhang, Chuo- Ling Chang, Ming Guang Yong, Juhyun Lee, et al. Mediapipe: A framework for building perception pipelines. arXiv preprint arXiv:1906.08172, 2019

    Pith/arXiv arXiv 1906

  72. [72]

    Alphapose: Whole-body regional multi-person pose estimation and tracking in real-time

    Hao-Shu Fang, Jiefeng Li, Hongyang Tang, Chao Xu, Haoyi Zhu, Yuliang Xiu, Yong-Lu Li, and Cewu Lu. Alphapose: Whole-body regional multi-person pose estimation and tracking in real-time. IEEE transactions on pattern analysis and machine intelligence, 45(6):7157–7173, 2022

    2022

  73. [73]

    2d human pose estimation: New benchmark and state of the art analysis

    Mykhaylo Andriluka, Leonid Pishchulin, Peter Gehler, and Bernt Schiele. 2d human pose estimation: New benchmark and state of the art analysis. In Proceedings of the IEEE Conference on computer Vision and Pattern Recognition , pages 3686–3693, 2014

    2014

  74. [74]

    Catalin Ionescu, Dragos Papava, Vlad Olaru, and Cristian Sminchisescu. Human3. 6m: Large scale datasets and predictive methods for 3d human sensing in natural environments. IEEE transactions on pattern analysis and machine intelligence, 36(7):1325–1339, 2013

    2013

  75. [75]

    Monocular 3d human pose estimation in the wild using improved cnn supervision

    Dushyant Mehta, Helge Rhodin, Dan Casas, Pascal Fua, Oleksandr Sotnychenko, Weipeng Xu, and Christian Theobalt. Monocular 3d human pose estimation in the wild using improved cnn supervision. In 2017 international conference on 3D vision (3DV), pages 506–516. IEEE, 2017

    2017

  76. [76]

    Flexible camera calibration by viewing a plane from unknown orientations

    Zhengyou Zhang. Flexible camera calibration by viewing a plane from unknown orientations. In Proceedings of the seventh ieee international conference on computer vision , volume 1, pages 666–673. Ieee, 1999

    1999