Validation of MediaPipe Pose for Sprint Time Measurement at 5, 10, and 20 Meters Against a Photocell System

Validation of MediaPipe Pose for Sprint Time Measurement at 5, 10, and 20 Meters Against a Photocell System | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Validation of MediaPipe Pose for Sprint Time Measurement at 5, 10, and 20 Meters Against a Photocell System Nobuchika Yamaki, Tenna Churiki This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7513017/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Accurate measurement of sprint time is essential in sports science and performance monitoring. Photocell timing systems are widely regarded as the gold standard, but they are costly and require specialized equipment. Recent advances in computer vision, such as pose estimation frameworks like OpenPose and MediaPipe Pose, enable motion analysis from standard smartphone video recordings. This study aimed to validate MediaPipe-based sprint time measurement against a photocell system across three distances commonly used in field testing: 5 m, 10 m, and 20 m. Twenty physically active adults performed repeated sprints, which were simultaneously recorded using photocells and a smartphone camera. Virtual start and finish lines were drawn on the video frame, and sprint time was defined as the frame difference between line-crossing events of body landmarks. Results demonstrated strong validity, with correlations above 0.95 for all distances. Mean absolute error (MAE) was highest for 5 m sprints (0.04 ± 0.06 s, ≈3% of total time), but decreased at 10 m (0.03 ± 0.05 s, ≈1.5%) and 20 m (0.03 ± 0.05 s, ≈1%). Bland–Altman analysis indicated minimal systematic bias. These findings suggest that MediaPipe provides a feasible, low-cost alternative for sprint timing, particularly reliable for distances ≥10 m, while caution is advised for very short sprints. Physical sciences/Engineering Physical sciences/Mathematics and computing Figures Figure 1 Figure 2 Introduction Sprint performance is a critical determinant of athletic success in multiple sports [ 1 ]. Precise measurement of sprint time is necessary for evaluating training effects, monitoring athlete progression, and conducting performance assessments in applied settings. Photocell timing systems are widely considered the gold standard for sprint evaluation because they detect the instant an athlete crosses an infrared beam [ 2 ]. However, these systems are costly, require stable setups, and are not always practical outside laboratory or specialized training environments. Computer vision techniques have recently enabled markerless analysis of human motion from standard video recordings [ 3 ]. Pose estimation frameworks such as OpenPose [ 4 ] and MediaPipe Pose [ 5 ] provide real-time detection of anatomical landmarks, and have been validated in tasks such as gait analysis, jump height estimation, and posture monitoring [ 3 , 6 ]. Extending this approach to sprint timing offers the potential for accessible, low-cost testing using only a smartphone camera. Although some studies have investigated video-based measurement of sprint parameters [ 6 ], few have validated computer vision–based methods across different sprint distances. Distances of 5 m, 10 m, and 20 m are widely used in sports testing, representing short acceleration, early acceleration, and near-maximal sprint phases, respectively [ 1 ]. Shorter distances are especially sensitive to measurement error, as even a small absolute error represents a large relative difference. The purpose of this study was to validate MediaPipe-based sprint timing against a photocell system at 5 m, 10 m, and 20 m. We hypothesized that MediaPipe-derived sprint times would show high correlations with photocell data across all distances, but relative error would be greater for shorter sprints. Methods Participants Twenty healthy adults (10 males, 10 females; age: 21.5 ± 2.3 years; body mass: 68.2 ± 9.4 kg) volunteered. Inclusion criteria included no musculoskeletal injuries in the previous six months and engagement in physical activity at least three times per week. Participation was entirely voluntary, and participants were informed that they could withdraw from the study at any time without penalty or explanation. All participants provided written informed consent before participation. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of TNQ Tech Co., Ltd. (Protocol Code: 2025-021, Date of Approval: 1 March, 2025). Sample Size Justification The required sample size was estimated a priori. For the primary analysis (convergent validity), we expected a large correlation (r = 0.90) between MediaPipe and photocell sprint times based on previous video-based validation studies [ 6 ]. Using G*Power 3.1 [ 9 ] with a two-tailed test, α = 0.05, and power (1–β) = 0.80, the minimum required sample size was n = 17. For reliability assessment using intraclass correlation coefficients (ICC), Walter et al. (1998) recommend at least 15–20 subjects when targeting an ICC ≥ 0.90 with α = 0.05 and power = 0.80 [ 10 ]. Therefore, we recruited 20 participants, which exceeds both criteria and ensures sufficient statistical power for correlation, ICC, and Bland–Altman analyses. Experimental Protocol Participants completed six maximal sprint trials covering 20 m on an outdoor synthetic track. Photocell gates (TCi PhotoGate Set, Brower Timing Systems, USA) were placed at 0 m (start), 5 m, 10 m, and 20 m. A smartphone camera (iPhone 14, 60 fps) was positioned laterally to the sprint lane at a perpendicular distance of approximately 10 m, with the optical axis aligned to the 10-m mark. The camera height was set to 1.2 m above ground level, mounted on a tripod, to capture the entire sprint distance in the sagittal plane. Calibration poles were placed at each photocell line to facilitate alignment of virtual start and finish lines during video analysis. Video Processing and MediaPipe Analysis Video recordings were processed using MediaPipe Pose [ 5 ]. Virtual start, 5 m, 10 m, and 20 m finish lines were drawn on the video frame, aligned with photocell beam positions using calibration poles. Sprint time was defined as the difference in frames between the line-crossing events of anatomical landmarks, divided by the frame rate (60 Hz). Start detection : The first frame in which the lead ankle (landmark 27/28) or torso (landmark 23/24) crossed the virtual start line. Split detection (5 m and 10 m) : Frames in which the torso (hip landmarks 23/24, averaged) crossed the respective virtual lines. Finish detection (20 m) : The frame in which the averaged hip landmarks crossed the virtual 20-m line. Hip landmarks were selected for intermediate and finish detections because they approximate the body’s center of mass and are less affected by arm or head swing. To reduce frame-to-frame jitter, landmark trajectories were smoothed using a three-frame moving average filter. All analyses were automated via a custom Python pipeline (Supplementary Code S1). Statistical Analysis MediaPipe and photocell times were compared at 5 m, 10 m, and 20 m. Agreement was evaluated using: Pearson correlation coefficients (r) Mean absolute error (MAE) and root mean square error (RMSE) Bland–Altman plots with 95% limits of agreement Intraclass correlation coefficients (ICC 2,1) for trial-to-trial reliability Statistical analyses were conducted in Python 3.12 using NumPy and SciPy. Significance was set at p < 0.05. Results All 120 trials were successfully recorded and analyzed. Summary statistics for each distance are presented in Table 1 . Raw paired times are provided in Supplementary Dataset S1. 5 m split : Photocell = 1.43 ± 0.17 s; MediaPipe = 1.47 ± 0.18 s. Correlation r = 0.93 (p < 0.001). MAE = 0.06 s, RMSE = 0.08 s. Bias = 0.04 s, 95% LoA = − 0.08 to 0.17 s. 10 m split : Photocell = 2.37 ± 0.24 s; MediaPipe = 2.39 ± 0.25 s. Correlation r = 0.98 (p < 0.001). MAE = 0.04 s, RMSE = 0.05 s. Bias = 0.02 s, 95% LoA = − 0.07 to 0.11 s. 20 m finish : Photocell = 3.28 ± 0.28 s; MediaPipe = 3.31 ± 0.28 s. Correlation r = 0.98 (p < 0.001). MAE = 0.05 s, RMSE = 0.06 s. Bias = 0.03 s, 95% LoA = − 0.07 to 0.13 s. Bland–Altman analysis revealed small mean biases (0.02–0.04 s) with narrow limits of agreement, consistent across distances. Reliability was excellent, with ICC values ranging from 0.94 to 0.96. Table 1 Comparison of sprint times measured with photocell and MediaPipe Pose at 5 m, 10 m, and 20 m. Values are presented as mean ± SD. Agreement is expressed by correlation coefficient (r), mean absolute error (MAE), root mean square error (RMSE), mean bias, and 95% limits of agreement (LoA). Distance Photocell (s) MediaPipe (s) Correlation (r) MAE (s) RMSE (s) Bias (s) 95% LoA (s) 5 m 1.43 ± 0.17 1.47 ± 0.18 0.93 0.06 0.08 0.04 -0.08 to 0.17 10 m 2.37 ± 0.24 2.39 ± 0.25 0.98 0.04 0.05 0.02 -0.07 to 0.11 20 m 3.28 ± 0.28 3.31 ± 0.28 0.98 0.05 0.06 0.03 -0.07 to 0.13 Discussion This study demonstrated that MediaPipe Pose provides valid sprint time measurements across multiple distances (5 m, 10 m, and 20 m) compared with photocell systems. Correlations were high at all distances (r = 0.93–0.98), and absolute errors were small (MAE = 0.04–0.06 s). Importantly, relative error was greatest at 5 m, where sprint duration is short. Here, an error of ≈ 0.06 s represented about 3–4% of total sprint time, which may influence interpretation in high-performance contexts. By contrast, relative errors at 10 m and 20 m were much smaller (≈ 1–2%), indicating stronger practical reliability for longer sprints. This distance-dependent pattern highlights that MediaPipe-based timing is especially well suited for assessments beyond very short acceleration phases. The methodological consistency with photocells represents a key strength. Both methods define sprint onset and completion as line-crossing events, reducing discrepancies that may arise from motion-onset definitions. Additionally, the use of hip landmarks for intermediate and finish detection provided robust measurements less influenced by arm or head movements. Nevertheless, limitations must be acknowledged. Accuracy is frame rate dependent; higher fps could reduce error, particularly at shorter distances. Landmark occlusion occasionally disrupted detection, though smoothing minimized this issue. Manual calibration of virtual lines may introduce slight alignment error, although calibration against photocell beams standardized the setup. In addition, the present validation was conducted only with healthy young adults on an outdoor track under favorable lighting conditions. The generalizability of these findings to different populations (e.g., elite athletes, older adults, or clinical cohorts), as well as to indoor environments or variable light conditions, remains to be confirmed. Furthermore, only sagittal-plane recordings were evaluated; oblique or frontal angles may reduce tracking accuracy and should be systematically tested in future work. Conclusion MediaPipe Pose provides valid sprint time measurement at 5 m, 10 m, and 20 m compared with photocell systems. While relative error is higher at 5 m, accuracy at 10 m and 20 m is sufficient for practical use. This low-cost, portable approach offers significant potential for field-based performance monitoring. Declarations Conflicts of Interest: Both authors are employees of TNQ Tech. However, this research was conducted independently and received no external or internal funding. The authors declare no other conflicts of interest. Funding: This research received no external funding. Author Contribution N.Y. wrote the manuscript, designed the study, and conducted the analysis. T.C. developed the code and assisted with figure creation. Both authors reviewed and approved the final version of the manuscript. Data Availability The data supporting the findings of this study are available as supplementary material attached to this submission. References Haugen, T. (2012). Sprint conditioning of elite soccer players: Sprint training and testing. Sports Medicine , 42(5), 403–416. https://doi.org/10.2165/11597020-000000000-00000 Cronin, J. B., & Templeton, R. (2008). Timing light height affects sprint times. Journal of Strength and Conditioning Research , 22(1), 318–320. https://doi.org/10.1519/JSC.0b013e31815f2b2e Colyer, S. L., Evans, M., Cosker, D. P., & Salo, A. I. T. (2018). 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. https://doi.org/10.1186/s40798-018-0139-y Cao, Z., Hidalgo, G., Simon, T., Wei, S. E., & Sheikh, Y. (2021). OpenPose: Realtime multi-person 2D pose estimation using part affinity fields. IEEE Transactions on Pattern Analysis and Machine Intelligence , 43(1), 172–186. https://doi.org/10.1109/TPAMI.2019.2929257 Lugaresi, C., Tang, J., Nash, H., McClanahan, C., et al. (2019). MediaPipe: A framework for building perception pipelines. arXiv preprint , arXiv:1906.08172. https://arxiv.org/abs/1906.08172 Nagahara, R., Mizutani, M., Matsuo, A., Kanehisa, H., & Fukunaga, T. (2021). Validity of spatiotemporal gait parameters during sprint acceleration using video-based tracking systems. Sports Biomechanics , 20(5), 550–562. https://doi.org/10.1080/14763141.2019.1584234 Clark, K. P., & Weyand, P. G. (2014). Are running speeds maximized with simple-spring stance mechanics? Journal of Applied Physiology , 117(6), 604–615. https://doi.org/10.1152/japplphysiol.00174.2014 Bezodis, N. E., North, J. S., & Razavet, J. L. (2017). Alterations to the orientation of the ground reaction force vector affect sprint acceleration performance in team sports athletes. Journal of Sports Sciences , 35(17), 1740–1747. https://doi.org/10.1080/02640414.2016.1239024 Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods , 39(2), 175–191. https://doi.org/10.3758/BF03193146 Walter, S. D., Eliasziw, M., & Donner, A. (1998). Sample size and optimal designs for reliability studies. Statistics in Medicine , 17(1), 101–110. https://doi.org/10.1002/(SICI)1097-0258(19980115)17:13.0.CO;2-E Additional Declarations No competing interests reported. Supplementary Files SupplementaryCodeS1.docx SupplementaryDatasetSprintTimes.csv Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7513017","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":509838434,"identity":"00df2741-41f4-4d74-ab17-eeebcdd15b50","order_by":0,"name":"Nobuchika Yamaki","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEElEQVRIiWNgGAWjYFACHgYGxgYJAygvgYcfxiJei2QbkDpAWAsDXAuDwTECWszZe49J/NxhYczfwGP46UZNmozx/d6Dtz9U2OUxsDcffIBFi2XPuTTJ3jMSZhIHeIylc47l8Jgd40u2OHAmuZiB51iyARYtBjdyzKQZ2yRsGA7wGEjnsFUAtfCYSRxsO5DYIJFjJoFPizzQlt85/yp4jNtAWv4R1mJmcIDHTDq3LYfHgA2kpQGPljPnki172ySMDQ+zlVnn9qXxSBzLMbY4cyw5sQ2XX473Hrzxs63OcN7x5s23c74l2/M3nzG8UVFjl9iPI8QQgJkDYSTYPWx4lYMBO8JIbF4YBaNgFIyCkQsAmIthmvIjDLMAAAAASUVORK5CYII=","orcid":"","institution":"TNQ Tech","correspondingAuthor":true,"prefix":"","firstName":"Nobuchika","middleName":"","lastName":"Yamaki","suffix":""},{"id":509838435,"identity":"f2052f93-6725-4a49-aca6-f2c4ba2c80f7","order_by":1,"name":"Tenna Churiki","email":"","orcid":"","institution":"TNQ Tech","correspondingAuthor":false,"prefix":"","firstName":"Tenna","middleName":"","lastName":"Churiki","suffix":""}],"badges":[],"createdAt":"2025-09-02 03:53:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7513017/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7513017/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":90571751,"identity":"601b7676-13a2-421f-b01f-3f85fdc79556","added_by":"auto","created_at":"2025-09-04 08:37:31","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":61100,"visible":true,"origin":"","legend":"\u003cp\u003eScatter plots comparing sprint times measured with MediaPipe Pose and the photocell system at 5 m, 10 m, and 20 m. Each plot shows a strong correlation between methods (r = 0.93–0.98). The dashed line represents the line of identity.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7513017/v1/fc2fac2beba99e77d2b57995.png"},{"id":90572686,"identity":"575f28c9-1d8c-498b-85b5-03bea4eb231c","added_by":"auto","created_at":"2025-09-04 08:45:31","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":77821,"visible":true,"origin":"","legend":"\u003cp\u003eBland–Altman plots showing agreement between MediaPipe and photocell sprint times at 5 m, 10 m, and 20 m. Mean biases were small (0.02–0.04 s), with narrow 95% limits of agreement. The dashed black line indicates mean bias, and the dashed grey lines represent the 95% limits of agreement (±1.96 SD).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7513017/v1/27ecaf348f0be2f6b6db5b77.png"},{"id":98623178,"identity":"b1ce5bac-6458-4494-a3da-c782eff5a6b9","added_by":"auto","created_at":"2025-12-19 17:05:03","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":552643,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7513017/v1/6f6250fe-e0e0-49f4-84a4-084c743f7d17.pdf"},{"id":90571756,"identity":"4d2947e8-ecbb-4b36-ab35-4238e48f5cd6","added_by":"auto","created_at":"2025-09-04 08:37:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":18490,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryCodeS1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7513017/v1/abb55901eeda818a66db7c5c.docx"},{"id":90571754,"identity":"af7dbf43-e639-4f65-bc38-0eab4c7d6646","added_by":"auto","created_at":"2025-09-04 08:37:32","extension":"csv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":3987,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryDatasetSprintTimes.csv","url":"https://assets-eu.researchsquare.com/files/rs-7513017/v1/2fb241351a727207e38a453f.csv"}],"financialInterests":"No competing interests reported.","formattedTitle":"Validation of MediaPipe Pose for Sprint Time Measurement at 5, 10, and 20 Meters Against a Photocell System","fulltext":[{"header":"Introduction","content":"\u003cp\u003eSprint performance is a critical determinant of athletic success in multiple sports [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Precise measurement of sprint time is necessary for evaluating training effects, monitoring athlete progression, and conducting performance assessments in applied settings. Photocell timing systems are widely considered the gold standard for sprint evaluation because they detect the instant an athlete crosses an infrared beam [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. However, these systems are costly, require stable setups, and are not always practical outside laboratory or specialized training environments.\u003c/p\u003e\u003cp\u003eComputer vision techniques have recently enabled markerless analysis of human motion from standard video recordings [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Pose estimation frameworks such as OpenPose [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] and MediaPipe Pose [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] provide real-time detection of anatomical landmarks, and have been validated in tasks such as gait analysis, jump height estimation, and posture monitoring [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Extending this approach to sprint timing offers the potential for accessible, low-cost testing using only a smartphone camera.\u003c/p\u003e\u003cp\u003eAlthough some studies have investigated video-based measurement of sprint parameters [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e], few have validated computer vision\u0026ndash;based methods across different sprint distances. Distances of 5 m, 10 m, and 20 m are widely used in sports testing, representing short acceleration, early acceleration, and near-maximal sprint phases, respectively [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Shorter distances are especially sensitive to measurement error, as even a small absolute error represents a large relative difference.\u003c/p\u003e\u003cp\u003eThe purpose of this study was to validate MediaPipe-based sprint timing against a photocell system at 5 m, 10 m, and 20 m. We hypothesized that MediaPipe-derived sprint times would show high correlations with photocell data across all distances, but relative error would be greater for shorter sprints.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eParticipants\u003c/h2\u003e\u003cp\u003eTwenty healthy adults (10 males, 10 females; age: 21.5\u0026thinsp;\u0026plusmn;\u0026thinsp;2.3 years; body mass: 68.2\u0026thinsp;\u0026plusmn;\u0026thinsp;9.4 kg) volunteered. Inclusion criteria included no musculoskeletal injuries in the previous six months and engagement in physical activity at least three times per week.\u003c/p\u003e\u003cp\u003eParticipation was entirely voluntary, and participants were informed that they could withdraw from the study at any time without penalty or explanation. All participants provided written informed consent before participation. The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of TNQ Tech Co., Ltd. (Protocol Code: 2025-021, Date of Approval: 1 March, 2025).\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eSample Size Justification\u003c/h3\u003e\n\u003cp\u003eThe required sample size was estimated a priori. For the primary analysis (convergent validity), we expected a large correlation (r\u0026thinsp;=\u0026thinsp;0.90) between MediaPipe and photocell sprint times based on previous video-based validation studies [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Using G*Power 3.1 [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] with a two-tailed test, α\u0026thinsp;=\u0026thinsp;0.05, and power (1\u0026ndash;β)\u0026thinsp;=\u0026thinsp;0.80, the minimum required sample size was n\u0026thinsp;=\u0026thinsp;17.\u003c/p\u003e\u003cp\u003eFor reliability assessment using intraclass correlation coefficients (ICC), Walter et al. (1998) recommend at least 15\u0026ndash;20 subjects when targeting an ICC\u0026thinsp;\u0026ge;\u0026thinsp;0.90 with α\u0026thinsp;=\u0026thinsp;0.05 and power\u0026thinsp;=\u0026thinsp;0.80 [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Therefore, we recruited 20 participants, which exceeds both criteria and ensures sufficient statistical power for correlation, ICC, and Bland\u0026ndash;Altman analyses.\u003c/p\u003e\n\u003ch3\u003eExperimental Protocol\u003c/h3\u003e\n\u003cp\u003eParticipants completed six maximal sprint trials covering 20 m on an outdoor synthetic track. Photocell gates (TCi PhotoGate Set, Brower Timing Systems, USA) were placed at 0 m (start), 5 m, 10 m, and 20 m. A smartphone camera (iPhone 14, 60 fps) was positioned laterally to the sprint lane at a perpendicular distance of approximately 10 m, with the optical axis aligned to the 10-m mark. The camera height was set to 1.2 m above ground level, mounted on a tripod, to capture the entire sprint distance in the sagittal plane. Calibration poles were placed at each photocell line to facilitate alignment of virtual start and finish lines during video analysis.\u003c/p\u003e\n\u003ch3\u003eVideo Processing and MediaPipe Analysis\u003c/h3\u003e\n\u003cp\u003eVideo recordings were processed using MediaPipe Pose [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Virtual start, 5 m, 10 m, and 20 m finish lines were drawn on the video frame, aligned with photocell beam positions using calibration poles. Sprint time was defined as the difference in frames between the line-crossing events of anatomical landmarks, divided by the frame rate (60 Hz).\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eStart detection\u003c/b\u003e: The first frame in which the lead ankle (landmark 27/28) or torso (landmark 23/24) crossed the virtual start line.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eSplit detection (5 m and 10 m)\u003c/b\u003e: Frames in which the torso (hip landmarks 23/24, averaged) crossed the respective virtual lines.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eFinish detection (20 m)\u003c/b\u003e: The frame in which the averaged hip landmarks crossed the virtual 20-m line.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eHip landmarks were selected for intermediate and finish detections because they approximate the body\u0026rsquo;s center of mass and are less affected by arm or head swing. To reduce frame-to-frame jitter, landmark trajectories were smoothed using a three-frame moving average filter.\u003c/p\u003e\u003cp\u003eAll analyses were automated via a custom Python pipeline (Supplementary Code S1).\u003c/p\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eMediaPipe and photocell times were compared at 5 m, 10 m, and 20 m. Agreement was evaluated using:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003ePearson correlation coefficients (r)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eMean absolute error (MAE) and root mean square error (RMSE)\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eBland\u0026ndash;Altman plots with 95% limits of agreement\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIntraclass correlation coefficients (ICC 2,1) for trial-to-trial reliability\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eStatistical analyses were conducted in Python 3.12 using NumPy and SciPy. Significance was set at p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eAll 120 trials were successfully recorded and analyzed. Summary statistics for each distance are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Raw paired times are provided in Supplementary Dataset S1.\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003e5 m split\u003c/b\u003e: Photocell\u0026thinsp;=\u0026thinsp;1.43\u0026thinsp;\u0026plusmn;\u0026thinsp;0.17 s; MediaPipe\u0026thinsp;=\u0026thinsp;1.47\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18 s. Correlation r\u0026thinsp;=\u0026thinsp;0.93 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). MAE\u0026thinsp;=\u0026thinsp;0.06 s, RMSE\u0026thinsp;=\u0026thinsp;0.08 s. Bias\u0026thinsp;=\u0026thinsp;0.04 s, 95% LoA = \u0026minus;\u0026thinsp;0.08 to 0.17 s.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003e10 m split\u003c/b\u003e: Photocell\u0026thinsp;=\u0026thinsp;2.37\u0026thinsp;\u0026plusmn;\u0026thinsp;0.24 s; MediaPipe\u0026thinsp;=\u0026thinsp;2.39\u0026thinsp;\u0026plusmn;\u0026thinsp;0.25 s. Correlation r\u0026thinsp;=\u0026thinsp;0.98 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). MAE\u0026thinsp;=\u0026thinsp;0.04 s, RMSE\u0026thinsp;=\u0026thinsp;0.05 s. Bias\u0026thinsp;=\u0026thinsp;0.02 s, 95% LoA = \u0026minus;\u0026thinsp;0.07 to 0.11 s.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003e20 m finish\u003c/b\u003e: Photocell\u0026thinsp;=\u0026thinsp;3.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.28 s; MediaPipe\u0026thinsp;=\u0026thinsp;3.31\u0026thinsp;\u0026plusmn;\u0026thinsp;0.28 s. Correlation r\u0026thinsp;=\u0026thinsp;0.98 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). MAE\u0026thinsp;=\u0026thinsp;0.05 s, RMSE\u0026thinsp;=\u0026thinsp;0.06 s. Bias\u0026thinsp;=\u0026thinsp;0.03 s, 95% LoA = \u0026minus;\u0026thinsp;0.07 to 0.13 s.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e Bland\u0026ndash;Altman analysis revealed small mean biases (0.02\u0026ndash;0.04 s) with narrow limits of agreement, consistent across distances. Reliability was excellent, with ICC values ranging from 0.94 to 0.96.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eComparison of sprint times measured with photocell and MediaPipe Pose at 5 m, 10 m, and 20 m. Values are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD. Agreement is expressed by correlation coefficient (r), mean absolute error (MAE), root mean square error (RMSE), mean bias, and 95% limits of agreement (LoA).\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"8\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDistance\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePhotocell (s)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMediaPipe (s)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCorrelation (r)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMAE (s)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eRMSE (s)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eBias (s)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003e95% LoA (s)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003e5 m\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e1.43\u0026thinsp;\u0026plusmn;\u0026thinsp;0.17\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e1.47\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.93\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.06\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.08\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.04\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e-0.08 to 0.17\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003e10 m\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e2.37\u0026thinsp;\u0026plusmn;\u0026thinsp;0.24\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e2.39\u0026thinsp;\u0026plusmn;\u0026thinsp;0.25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.04\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.05\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.02\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e-0.07 to 0.11\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003e20 m\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e\u003cp\u003e3.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e\u003cp\u003e3.31\u0026thinsp;\u0026plusmn;\u0026thinsp;0.28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.98\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.05\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.06\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.03\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e-0.07 to 0.13\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study demonstrated that MediaPipe Pose provides valid sprint time measurements across multiple distances (5 m, 10 m, and 20 m) compared with photocell systems. Correlations were high at all distances (r\u0026thinsp;=\u0026thinsp;0.93\u0026ndash;0.98), and absolute errors were small (MAE\u0026thinsp;=\u0026thinsp;0.04\u0026ndash;0.06 s).\u003c/p\u003e\u003cp\u003eImportantly, relative error was greatest at 5 m, where sprint duration is short. Here, an error of \u0026asymp;\u0026thinsp;0.06 s represented about 3\u0026ndash;4% of total sprint time, which may influence interpretation in high-performance contexts. By contrast, relative errors at 10 m and 20 m were much smaller (\u0026asymp;\u0026thinsp;1\u0026ndash;2%), indicating stronger practical reliability for longer sprints. This distance-dependent pattern highlights that MediaPipe-based timing is especially well suited for assessments beyond very short acceleration phases.\u003c/p\u003e\u003cp\u003eThe methodological consistency with photocells represents a key strength. Both methods define sprint onset and completion as line-crossing events, reducing discrepancies that may arise from motion-onset definitions. Additionally, the use of hip landmarks for intermediate and finish detection provided robust measurements less influenced by arm or head movements.\u003c/p\u003e\u003cp\u003eNevertheless, limitations must be acknowledged. Accuracy is frame rate dependent; higher fps could reduce error, particularly at shorter distances. Landmark occlusion occasionally disrupted detection, though smoothing minimized this issue. Manual calibration of virtual lines may introduce slight alignment error, although calibration against photocell beams standardized the setup.\u003c/p\u003e\u003cp\u003eIn addition, the present validation was conducted only with healthy young adults on an outdoor track under favorable lighting conditions. The generalizability of these findings to different populations (e.g., elite athletes, older adults, or clinical cohorts), as well as to indoor environments or variable light conditions, remains to be confirmed. Furthermore, only sagittal-plane recordings were evaluated; oblique or frontal angles may reduce tracking accuracy and should be systematically tested in future work.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eMediaPipe Pose provides valid sprint time measurement at 5 m, 10 m, and 20 m compared with photocell systems. While relative error is higher at 5 m, accuracy at 10 m and 20 m is sufficient for practical use. This low-cost, portable approach offers significant potential for field-based performance monitoring.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eConflicts of Interest:\u003c/h2\u003e\u003cp\u003eBoth authors are employees of TNQ Tech. However, this research was conducted independently and received no external or internal funding. The authors declare no other conflicts of interest.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding:\u003c/h2\u003e\u003cp\u003eThis research received no external funding.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eN.Y. wrote the manuscript, designed the study, and conducted the analysis. T.C. developed the code and assisted with figure creation. Both authors reviewed and approved the final version of the manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data supporting the findings of this study are available as supplementary material attached to this submission.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eHaugen, T. (2012). Sprint conditioning of elite soccer players: Sprint training and testing. \u003cem\u003eSports Medicine\u003c/em\u003e, 42(5), 403\u0026ndash;416.\u0026nbsp;https://doi.org/10.2165/11597020-000000000-00000\u003c/li\u003e\n \u003cli\u003eCronin, J. B., \u0026amp; Templeton, R. (2008). Timing light height affects sprint times. \u003cem\u003eJournal of Strength and Conditioning Research\u003c/em\u003e, 22(1), 318\u0026ndash;320.\u0026nbsp;https://doi.org/10.1519/JSC.0b013e31815f2b2e\u003c/li\u003e\n \u003cli\u003eColyer, S. L., Evans, M., Cosker, D. P., \u0026amp; Salo, A. I. T. (2018). A review of the evolution of vision-based motion analysis and the integration of advanced computer vision methods towards developing a markerless system. \u003cem\u003eSports Medicine - Open\u003c/em\u003e, 4(1), 24.\u0026nbsp;https://doi.org/10.1186/s40798-018-0139-y\u003c/li\u003e\n \u003cli\u003eCao, Z., Hidalgo, G., Simon, T., Wei, S. E., \u0026amp; Sheikh, Y. (2021). OpenPose: Realtime multi-person 2D pose estimation using part affinity fields. \u003cem\u003eIEEE Transactions on Pattern Analysis and Machine Intelligence\u003c/em\u003e, 43(1), 172\u0026ndash;186.\u0026nbsp;https://doi.org/10.1109/TPAMI.2019.2929257\u003c/li\u003e\n \u003cli\u003eLugaresi, C., Tang, J., Nash, H., McClanahan, C., et al. (2019). MediaPipe: A framework for building perception pipelines. \u003cem\u003earXiv preprint\u003c/em\u003e, arXiv:1906.08172.\u0026nbsp;https://arxiv.org/abs/1906.08172\u003c/li\u003e\n \u003cli\u003eNagahara, R., Mizutani, M., Matsuo, A., Kanehisa, H., \u0026amp; Fukunaga, T. (2021). Validity of spatiotemporal gait parameters during sprint acceleration using video-based tracking systems. \u003cem\u003eSports Biomechanics\u003c/em\u003e, 20(5), 550\u0026ndash;562.\u0026nbsp;https://doi.org/10.1080/14763141.2019.1584234\u003c/li\u003e\n \u003cli\u003eClark, K. P., \u0026amp; Weyand, P. G. (2014). Are running speeds maximized with simple-spring stance mechanics? \u003cem\u003eJournal of Applied Physiology\u003c/em\u003e, 117(6), 604\u0026ndash;615.\u0026nbsp;https://doi.org/10.1152/japplphysiol.00174.2014\u003c/li\u003e\n \u003cli\u003eBezodis, N. E., North, J. S., \u0026amp; Razavet, J. L. (2017). Alterations to the orientation of the ground reaction force vector affect sprint acceleration performance in team sports athletes. \u003cem\u003eJournal of Sports Sciences\u003c/em\u003e, 35(17), 1740\u0026ndash;1747.\u0026nbsp;https://doi.org/10.1080/02640414.2016.1239024\u003c/li\u003e\n \u003cli\u003eFaul, F., Erdfelder, E., Lang, A.-G., \u0026amp; Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. \u003cem\u003eBehavior Research Methods\u003c/em\u003e, 39(2), 175\u0026ndash;191.\u0026nbsp;https://doi.org/10.3758/BF03193146\u003c/li\u003e\n \u003cli\u003eWalter, S. D., Eliasziw, M., \u0026amp; Donner, A. (1998). Sample size and optimal designs for reliability studies. \u003cem\u003eStatistics in Medicine\u003c/em\u003e, 17(1), 101\u0026ndash;110. https://doi.org/10.1002/(SICI)1097-0258(19980115)17:1\u0026lt;101::AID-SIM727\u0026gt;3.0.CO;2-E\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7513017/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7513017/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate measurement of sprint time is essential in sports science and performance monitoring. Photocell timing systems are widely regarded as the gold standard, but they are costly and require specialized equipment. Recent advances in computer vision, such as pose estimation frameworks like OpenPose and MediaPipe Pose, enable motion analysis from standard smartphone video recordings. This study aimed to validate MediaPipe-based sprint time measurement against a photocell system across three distances commonly used in field testing: 5 m, 10 m, and 20 m. Twenty physically active adults performed repeated sprints, which were simultaneously recorded using photocells and a smartphone camera. Virtual start and finish lines were drawn on the video frame, and sprint time was defined as the frame difference between line-crossing events of body landmarks. Results demonstrated strong validity, with correlations above 0.95 for all distances. Mean absolute error (MAE) was highest for 5 m sprints (0.04 ± 0.06 s, ≈3% of total time), but decreased at 10 m (0.03 ± 0.05 s, ≈1.5%) and 20 m (0.03 ± 0.05 s, ≈1%). Bland–Altman analysis indicated minimal systematic bias. These findings suggest that MediaPipe provides a feasible, low-cost alternative for sprint timing, particularly reliable for distances ≥10 m, while caution is advised for very short sprints.\u003c/p\u003e","manuscriptTitle":"Validation of MediaPipe Pose for Sprint Time Measurement at 5, 10, and 20 Meters Against a Photocell System","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-04 08:37:27","doi":"10.21203/rs.3.rs-7513017/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"b2560d24-bea3-469b-b1dd-46a98185154d","owner":[],"postedDate":"September 4th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":54153243,"name":"Physical sciences/Engineering"},{"id":54153244,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2025-12-17T18:23:56+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-04 08:37:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7513017","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7513017","identity":"rs-7513017","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}