ArticlePDF Available

Is the high-speed camera-based method a plausible option for bar velocity assessment during resistance training?

Authors:

Abstract and Figures

This study aims to identify the measurement error associated with the mean movement velocity when using high-speed camera-based methods and video analysis during resistance training. Eleven resistance-trained men (26.0 ± 3.4 years) completed a progressive loading test in bench press exercise. Measurements from concentric mean velocity (MV), distance and time were obtained from a linear velocity transducer (T-Force) and videos recorded with high speed cameras on readily available smartphones (Samsung S6, Xiaomi A1, and iPhone X) and digital photo cameras (Casio FH20). Videos were examined using video analysis software (Kinovea). Despite the high correlations detected, the Bland-Altman analyses revealed that all high speed cameras produced substantial overestimation of barbell MV against high loads >60% 1RM (MV error = 0.06 ± 0.03 m·s-1 to 0.08 ± 0.04 m·s-1), but mainly against low loads <60% 1RM (MV error = 0.13 ± 0.06 m·s-1 to 0.20 ± 0.09 m·s-1). The maximum estimation error of the load being lifted (%1RM) was considerable both for low (8.5% to 12.7% 1RM) and high loads (13.9% to 22.6% 1RM). Among other practical limitations, the video-based system using different high-speed cameras and smartphone devices presents severe limitations when estimating mean concentric velocity, especially when recording low loads at high velocity.
Content may be subject to copyright.
Is the high-speed camera-based method a plausible option for bar
velocity assessment during resistance training?
Alejandro Sánchez-Pay, Javier Courel-Ibáñez, Alejandro Martínez-Cava, Elena Conesa-Ros,
Ricardo Morán-Navarro, Jesús G. Pallarés
Human Performance and Sports Science Laboratory, Faculty of Sport Sciences, University of Murcia, Spain
article info
Article history:
Received 11 September 2018
Received in revised form 3 January 2019
Accepted 5 January 2019
Available online 10 January 2019
Keywords:
Velocity-based resistance training
Strength
Powerlifting
Load monitoring
Validity analysis
abstract
This study aims to identify the measurement error associated with the mean movement velocity when
using high-speed camera-based methods and video analysis during resistance training. Eleven
resistance-trained men (26.0 ± 3.4 years) completed a progressive loading test in bench press exercise.
Measurements from concentric mean velocity (MV), distance and time were obtained from a linear veloc-
ity transducer (T-Force) and videos recorded with high speed cameras on readily available smartphones
(Samsung S6, Xiaomi A1, and iPhone X) and digital photo cameras (Casio FH20). Videos were examined
using video analysis software (Kinovea). Despite the high correlations detected, the Bland-Altman anal-
yses revealed that all high speed cameras produced substantial overestimation of barbell MV against high
loads >60% 1RM (MV error = 0.06 ± 0.03 ms
1
to 0.08 ± 0.04 ms
1
), but mainly against low loads <60%
1RM (MV error = 0.13 ± 0.06 ms
1
to 0.20 ± 0.09 ms
1
). The maximum estimation error of the load being
lifted (%1RM) was considerable both for low (8.5–12.7% 1RM) and high loads (13.9–22.6% 1RM). Among
other practical limitations, the video-based system using different high-speed cameras and smartphone
devices presents severe limitations when estimating mean concentric velocity, especially when recording
low loads at high velocity.
Ó2019 Elsevier Ltd. All rights reserved.
1. Introduction
In recent years, sports scientists and coaches are witnessing a
tremendous growth in popularity of velocity-based resistance
training (VBRT) for quantifying force production and power output
during resistance training [1–3]. This approach, based on barbell
velocity and displacement, is a valid, reliable and highly sensitive
method to: (1) Determine an athlete’s maximum strength without
the need to perform one repetition maximum (1RM) or maximum
number of repetitions to failure (nRM) tests [2]; (2) Determine the
%1RM that is being used from the first repetition performed at
maximal voluntary velocity for a given load [4]; (3) Estimate the
muscle power output production [5]; and (4) Quantify the neuro-
muscular fatigue induced by resistance exercise using a non-
invasive and objective method [6–10]. Despite these advantages,
the VBRT requires the use of highly accurate and reliable technolo-
gies to be successfully implemented.
The use of accurate assessment tools is an essential component
of training at elite level, due to the fact that very small changes in
performance may largely impact on the athletes’ chances of win-
ning [11]. Given that minimal changes in the workload can induce
critical improvements on the neuromuscular and functional per-
formance in well-trained athletes, it seems critical to examine
the concurrent validity of the assessment protocol to minimise
the measurement error [12–14]. Regarding barbell velocity moni-
toring, several recent studies conducted on experienced athletes,
reported that minor increments in strength (2–5% 1RM) resulted
in very important performance enhancement (effect size = 0.20–
0.85) at different maximal and submaximal loading intensities
[1,15–17]. The meaningfulness of improvements is associated with
the resistance training program [1], the dehydration status [18],
the acute ingestion of ergogenic aids [17] or the circadian rhythm
effect [15]. The most recent studies around the VBRT approach sta-
ted that 0.07 ms
1
increments resulted in 5.0% 1RM enhance-
ments in the most common resistance training exercises as
bench press and squat [2,5,19].
These advantages and new practical applications have aroused
an interest in VBRT equipment and methods, increasing the num-
ber of available devices for quantifying the barbell velocity, as well
as investigations testing their validity and reliability [5,19,20].
Kinematic systems such as the linear position and linear velocity
https://doi.org/10.1016/j.measurement.2019.01.006
0263-2241/Ó2019 Elsevier Ltd. All rights reserved.
Corresponding author at: C/ Argentina, s/n, Santiago de la Ribera, Murcia, Spain.
E-mail address: jgpallares@um.es (J.G. Pallarés).
Measurement 137 (2019) 355–361
Contents lists available at ScienceDirect
Measurement
journal homepage: www.elsevier.com/locate/measurement
... However, their high cost and difficult portability, as well as the requirement to operate them using a computer or tablet makes their use at certain circumstances, not so easy for practical use. On the other hand, high speed video cameras (included in smartphones) in combination with kinematic software or smartphone applications are more affordable and easy-to use, while their video capturing systems are able of high speed recording (120-300 frames per secondfps) compared with the standard types of video cameras (30 fps), thus allowing for more accurate measurements [4,17,25,34]. ...
... Regarding the determination of training intensity with the use of VA, research has shown that a mean velocity bias from 0.13 m/s to 0.20 m/s could cause errors in the estimation of maximum strength (1RM) by 13.9% to 22.6% in the bench press exercise [34]. However, in our study the differences in mean movement velocity between the two measuring systems was small (0.06 m/s). ...
... Thus the error of estimation of maximum strength with the free software is limited to 6-7% [14]. Also, it has been shown that high speed video cameras (digital cameras or smart phone cameras) produced significant overestimation (>0.13 m/s) of barbell movement velocity especially at high velocities >0.8 m/s [34]. In the present study we used ballistic push offs, and the greater percentage of muscle actions was over 0.8 m/s (Fig. 1) however the difference in mean and peak velocity were very small (0.06 and 0.01 m/s, respectively) compared with the above previous studies. ...
Article
Full-text available
Study aim : This study compared movement velocity and force-velocity profile parameters measured by a free video analysis software program, with the use of a high-speed video recording, and a validated linear position transducer (LPT). Material and methods : Ten team-sports athletes performed double-leg and single-leg ballistic lower limb extensions on a leg press machine against a wide range of resistive loads. Each repetition was recorded by the LPT a high-speed camera (300 fps), and later analysed with a free video analysis software program. Results : Mean and peak movement velocity presented high reliability (ICC: 0.990 and 0.988, p < 0.001) and agreement between the two measuring systems (systematic bias: –0.06 ± 0.04 and –0.01 ± 0.03 m/s, respectively). Force-velocity profile parameters were also similar: maximum velocity at zero load (Vo: 1.79 ± 0.15 vs. 1.78 ± 0.12 m/s, p = 0.64), slope (b: –1585 ± 503 vs. –1562 ± 438 N · s/m, p = 0.43), maximum force at zero velocity (Fo: 2835 ± 937 vs. 2749 ± 694 N, p = 0.41) and maximum power (1274 ± 451 vs 1214 ± 285 W, p = 0.38). Both measuring systems could similarly detect the individual force or velocity deficit (p=0.91). Conclusion : In conclusion, a free video analysis software combined with a high-speed camera was shown to be a reliable, accurate, low bias and cost-effective method in velocity-based testing.
... Furthermore, in recent times, there have been a range of new devices that monitor resistance training outputs, with these being made possible through advancements in technology [22]. Examples of these include optic laser devices and the cameras within smartphones [22,23]. While validity and reliability data have been published on these new devices, they have sparingly been compared to linear transducer (i.e., either LPTs or LVTs) and accelerometer data [24]. ...
... Balsalobre-Fernández et al. [69], Mitter et al. [9], Perez-Castilla et al. [10], Thompson et al. [24] Balsalobre-Fernández et al. [69], Perez-Castilla et al. [10], Thompson et al. [24] Bar Sensei Beckham et al. [14], Thompson et al. [24], Abbott et al. [59] Beckham et al. [14], Thompson Kinovea via iPhone X Sanchez-Pay et al. [23] Kinovea via Casio FH20 ...
... Sanchez-Pay et al. [23] Validity and Reliability of Resistance Training Devices ...
Article
Full-text available
Background Monitoring resistance training has a range of unique difficulties due to differences in physical characteristics and capacity between athletes, and the indoor environment in which it often occurs. Traditionally, methods such as volume load have been used, but these have inherent flaws. In recent times, numerous portable and affordable devices have been made available that purport to accurately and reliably measure kinetic and kinematic outputs, potentially offering practitioners a means of measuring resistance training loads with confidence. However, a thorough and systematic review of the literature describing the reliability and validity of these devices has yet to be undertaken, which may lead to uncertainty from practitioners on the utility of these devices. Objective A systematic review of studies that investigate the validity and/or reliability of commercially available devices that quantify kinetic and kinematic outputs during resistance training. Methods Following PRISMA guidelines, a systematic search of SPORTDiscus, Web of Science, and Medline was performed; studies included were (1) original research investigations; (2) full-text articles written in English; (3) published in a peer-reviewed academic journal; and (4) assessed the validity and/or reliability of commercially available portable devices that quantify resistance training exercises. Results A total of 129 studies were retrieved, of which 47 were duplicates. The titles and abstracts of 82 studies were screened and the full text of 40 manuscripts were assessed. A total of 31 studies met the inclusion criteria. Additional 13 studies, identified via reference list assessment, were included. Therefore, a total of 44 studies were included in this review. Conclusion Most of the studies within this review did not utilise a gold-standard criterion measure when assessing validity. This has likely led to under or overreporting of error for certain devices. Furthermore, studies that have quantified intra-device reliability have often failed to distinguish between technological and biological variability which has likely altered the true precision of each device. However, it appears linear transducers which have greater accuracy and reliability compared to other forms of device. Future research should endeavour to utilise gold-standard criterion measures across a broader range of exercises (including weightlifting movements) and relative loads.
... For this purpose, different commercial devices can be used to quantify velocity [23]. Among the available options, solutions can be grouped as follows [24]: (i) isoinertial dynamometers consisting of a cable-extension linear velocity transducer attached to the barbell [25][26][27], (ii) optical motion sensing systems or optoelectronic systems [28][29][30][31], (iii) smartphone applications involving frame-by-frame manual inspections [29,32,33], and (iv) inertial measurement units (IMUs) [34]. Since these different technologies offer different possibilities, it can be considered that IMUs represent the most easy-to-use solution because no cable-extension is needed-the sensor simply needs to be attached to the barbell. ...
Article
Full-text available
The use of inertial measurement unit (IMU) has become popular in sports assessment. In the case of velocity-based training (VBT), there is a need to measure barbell velocity in each repetition. The use of IMUs may make the monitoring process easier; however, its validity and reliability should be established. Thus, this systematic review aimed to (1) identify and summarize studies that have examined the validity of wearable wireless IMUs for measuring barbell velocity and (2) identify and summarize studies that have examined the reliability of IMUs for measuring barbell velocity. A systematic review of Cochrane Library, EBSCO, PubMed, Scielo, Scopus, SPORTDiscus, and Web of Science databases was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. From the 161 studies initially identified, 22 were fully reviewed, and their outcome measures were extracted and analyzed. Among the eight different IMU models, seven can be considered valid and reliable for measuring barbell velocity. The great majority of IMUs used for measuring barbell velocity in linear trajectories are valid and reliable, and thus can be used by coaches for external load monitoring.
... cm for 240 fps. Contrastingly, when the information retrieval is performed by human digitizing, including human errors due to observation, low video frame rates may pose severe limitations when estimating velocity outcomes at high velocity barbell displacement (>0.80 m/s) [52]. The automatic tracking of the proposed system avoids such human errors, so lower frame rates could be used, as with LPT encoders in [51]. ...
Article
Full-text available
Velocity-based training is a contemporary method used by sports coaches to prescribe the optimal loading based on the velocity of movement of a load lifted. The most employed and accurate instruments to monitor velocity are linear position transducers. Alternatively, smartphone apps compute mean velocity after each execution by manual on-screen digitizing, introducing human error. In this paper, a video-based instrument delivering unattended, real-time measures of barbell velocity with a smartphone high-speed camera has been developed. A custom image-processing algorithm allows for the detection of reference points of a multipower machine to autocalibrate and automatically track barbell markers to give real-time kinematic-derived parameters. Validity and reliability were studied by comparing the simultaneous measurement of 160 repetitions of back squat lifts executed by 20 athletes with the proposed instrument and a validated linear position transducer, used as a criterion. The video system produced practically identical range, velocity, force, and power outcomes to the criterion with low and proportional systematic bias and random errors. Our results suggest that the developed video system is a valid, reliable, and trustworthy instrument for measuring velocity and derived variables accurately with practical implications for use by coaches and practitioners.
... Hassan et al. [17] presented a feasibility study of the heart rate measurement in which they used a digital camera for health monitoring. Sánchez-Pay et al. [18] presented an article describing the measurement error identification associated with the mean movement velocity, in which methods based on a high-speed camera and video analysis were used during resistance experiments. ...
Article
Full-text available
The foil bearing consists of parts made of very thin, properly shaped foils. Usually, it is very difficult or even impossible to measure the vibrations of these elements during the bearing operation using traditional sensors. Therefore, the authors of this article have proposed an entirely new approach to this issue. This article discusses the analysis of vibrations of the structural supporting layer of a gas foil bearing at high rotational speeds. Instead of using a traditional method to measure the bearing journal movement, the measurement was performed using an ultra-high-speed digital camera. This type of measurement was used for the first time to analyse foil bearing displacement. It turned out that doing so can give a far more vibrant picture of what is happening in gas foil bearings during their operation. The article includes an analysis of foil vibrations. This phenomenon has already been analysed numerically, and this is the first time it has been analysed experimentally. The registered motion of the foils can be compared with the results obtained from numerical models, thus allowing their further development. One such comparison is shown in this article.
... Kinovea is free open-source motion analysis software under GPLv2 license, developed by sport and health professionals, programmers, researchers, and athletes in worldwide non-profit collaboration. In the field of sports, Kinovea has been used as a position-based instrument for measuring coordinates data and perspective [24], lower limb angle [40][41][42], bar velocity through tracking [43] or drop jump [44]. Similarly, as a time-based instrument, Kinovea has also been used to measure temporal parameters in jump height assessment, such as flight time [45][46][47]. ...
Article
Jumping is a simple exercise determined by several biomechanical and physiological factors. Measures of vertical jump height are common and easy to administer tests of lower limb muscle power that are carried out with several types of equipment. This study aimed to validate and address the usefulness of the combination of smartphone and computer-based applications (Smartphone-Kinovea) against a laboratory-based Motion Capture System. One hundred and twelve healthy adults performed three maximal-effort countermovement jumps each. Both instruments measured the heights of the 336 trials concurrently while tracking the excursion of the body center of gravity. The vertical velocity at take-off vto and the impulse J were computed with jump height h measures. Intraclass correlation coefficient (ICC) results indicated very high agreement for h and vto (0.985) and almost perfect agreement for J (0.997), and Cronbach's α=0.99. Low mean differences were observed between instruments for h: -0.22 ± 1.15 cm, vto: -0.01 ± 0.04 m/s, and J: -0.56 ± 2.92 Ns, all p<0.01. The smallest worthwhile change (SWC) and the typical error of measurement (SEM) were 1.34 cm, 0.81 cm for h; 1.15 m/s, 0.03 m/s for vto, and 2.93 Ns, 2.25 Ns for J, so the usefulness of the method is established (SWC/SEM>1). Bland-Altman plots showed very low mean systematic bias ± random errors (-0.22 ± 2.25 cm; -0.01 ± 0.08 m/s; -0.56 ± 5.73 Ns), without association between their magnitudes (r²=0.005, r²=0.005, r²=0.001). Finally, very high to practically perfect correlation between isntruments were observed (r= 0.985; r= 0.986; r= 0.997). Our results suggest that the Smartphone-Kinovea method is a valid and reliable, low-cost instrument to monitor changes in jump performance in a healthy, active population diverse in gender and physical condition.
... Despite the increasing popularity of smartphone apps for VBT, our results showed that the My Lift was the least reliable tool compared to the other available devices (Fig 5E and 5F). These results are consistent with previous research [19,27,37] suggesting that My Lift might be only reliable to track slow lifts with heavy loads. Although this might be useful to conduct routinely submaximal loading tests at > 80% 1RM and to avoid reaching the 100%, the use of the My Lift app for VBT is ill-advisable due to its large measurement errors (SEM > 0.10 m/s, SDC > 0.23 m/s) when tracking lifts > 0.30 m/ s, making it difficult to determine load adjustments or performance fluctuations with sufficient accuracy. ...
Article
Full-text available
This study investigated the inter- and intra-device agreement of four new devices marketed for barbell velocity measurement. Mean, mean propulsive and peak velocity outcomes were obtained for bench press and full squat exercises along the whole load-velocity spectrum (from light to heavy loads). Measurements were simultaneously registered by two linear velocity transducers T-Force, two linear position transducers Speed4Lifts, two smartphone video-based systems My Lift, and one 3D motion analysis system STT. Calculations included infraclass correlation coefficient (ICC), Bland-Altman Limits of Agreement (LoA), standard error of measurement (SEM), smallest detectable change (SDC) and maximum errors (MaxError). Results were reported in absolute (m/s) and relative terms (%1RM). Three velocity segments were differentiated according to the velocity-load relationships for each exercise: heavy (≥ 80% 1RM), medium (50% < 1RM < 80%) and light loads (≤ 50% 1RM). Criteria for acceptable reliability were ICC > 0.990 and SDC < 0.07 m/s (~5% 1RM). The T-Force device shown the best intra-device agreement (SDC = 0.01–0.02 m/s, LoA <0.01m/s, MaxError = 1.3–2.2%1RM). The Speed4Lifts and STT were found as highly reliable, especially against lifting velocities ≤1.0 m/s (Speed4Lifts, SDC = 0.01–0.05 m/s; STT, SDC = 0.02–0.04 m/s), whereas the My Lift app showed the worst results with errors well above the acceptable levels (SDC = 0.26–0.34 m/s, MaxError = 18.9–24.8%1RM). T-Force stands as the preferable option to assess barbell velocity and to identify technical errors of measurement for emerging monitoring technologies. Both the Speed4Lifts and STT are fine alternatives to T-Force for measuring velocity against high-medium loads (velocities ≤ 1.0 m/s), while the excessive errors of the newly updated My Lift app advise against the use of this tool for velocity-based resistance training.
... Despite the increasing popularity of smartphone apps for VBT, our results showed that the My Lift was the least reliable tool compared to the other available devices (Fig 5E and 5F). These results are consistent with previous research [19,27,37] suggesting that My Lift might be only reliable to track slow lifts with heavy loads. Although this might be useful to conduct routinely submaximal loading tests at > 80% 1RM and to avoid reaching the 100%, the use of the My Lift app for VBT is ill-advisable due to its large measurement errors (SEM > 0.10 m/s, SDC > 0.23 m/s) when tracking lifts > 0.30 m/ s, making it difficult to determine load adjustments or performance fluctuations with sufficient accuracy. ...
Article
Full-text available
This study investigated the inter- and intra-device agreement of four new devices marketed for barbell velocity measurement. Mean, mean propulsive and peak velocity outcomes were obtained for bench press and full squat exercises along the whole load-velocity spectrum (from light to heavy loads). Measurements were simultaneously registered by two linear velocity transducers T-Force, two linear position transducers Speed4Lifts, two smartphone video-based systems My Lift, and one 3D motion analysis system STT. Calculations included infraclass correlation coefficient (ICC), Bland-Altman Limits of Agreement (LoA), standard error of measurement (SEM), smallest detectable change (SDC) and maximum errors (MaxError). Results were reported in absolute (m/s) and relative terms (%1RM). Three velocity segments were differentiated according to the velocity-load relationships for each exercise: heavy (≥ 80% 1RM), medium (50% < 1RM < 80%) and light loads (≤ 50% 1RM). Criteria for acceptable reliability were ICC > 0.990 and SDC < 0.07 m/s (~5% 1RM). The T-Force device shown the best intra-device agreement (SDC = 0.01–0.02 m/s, LoA <0.01m/s, MaxError = 1.3–2.2%1RM). The Speed4Lifts and STT were found as highly reliable, especially against lifting velocities ≤1.0 m/s (Speed4Lifts, SDC = 0.01–0.05 m/s; STT, SDC = 0.02–0.04 m/s), whereas the My Lift app showed the worst results with errors well above the acceptable levels (SDC = 0.26–0.34 m/s, MaxError = 18.9–24.8%1RM). T-Force stands as the preferable option to assess barbell velocity and to identify technical errors of measurement for emerging monitoring technologies. Both the Speed4Lifts and STT are fine alternatives to T-Force for measuring velocity against high-medium loads (velocities ≤ 1.0 m/s), while the excessive errors of the newly updated My Lift app advise against the use of this tool for velocity-based resistance training.
Article
Full-text available
This study aimed to analyze the agreement between five bar velocity monitoring devices, currently used in resistance training, to determine the most reliable device based on reproducibility (between-device agreement for a given trial) and repeatability (between-trial variation for each device). Seventeen resistance-trained men performed duplicate trials against seven increasing loads (20-30-40-50-60-70-80 kg) while obtaining mean, mean propulsive and peak velocity outcomes in the bench press, full squat and prone bench pull exercises. Measurements were simultaneously registered by two linear velocity transducers (LVT), two linear position transducers (LPT), two optoelectronic camera-based systems (OEC), two smartphone video-based systems (VBS) and one accelerometer (ACC). A comprehensive set of statistics for assessing reliability was used. Magnitude of errors was reported both in absolute (m s⁻¹) and relative terms (%1RM), and included the smallest detectable change (SDC) and maximum errors (MaxError). LVT was the most reliable and sensitive device (SDC 0.02–0.06 m s⁻¹, MaxError 3.4–7.1% 1RM) and the preferred reference to compare with other technologies. OEC and LPT were the second-best alternatives (SDC 0.06–0.11 m s⁻¹), always considering the particular margins of error for each exercise and velocity outcome. ACC and VBS are not recommended given their substantial errors and uncertainty of the measurements (SDC > 0.13 m s⁻¹).
Article
Full-text available
This study investigated the validity and reliability of the GymAware PowerTool (GPT). Thirteen resistance trained participants completed three visits, consisting of three repetitions of free-weight back squat, bench press, deadlift (80% one repetition maximum), and countermovement jump. Bar displacement, peak and mean velocity, peak and mean force, and jump height were calculated using the GPT, a three-dimensional motion capture system (Motion Analysis Corporation; 150 Hz), and a force plate (Kistler; 1500 Hz). Least products regression were used to compare agreeability between devices. A within-trial one-way ANOVA, typical error (TE; %), and smallest worthwhile change (SWC) were used to assess reliability. Regression analysis resulted in R² values of >0.85 for all variables excluding deadlift mean velocity (R² = 0.54–0.69). Significant differences were observed between visits 3-2 for bench press bar displacement (0.395 ± 0.055 m; 0.383 ± 0.053 m), and deadlift bar displacement (0.557 ± 0.034 m; 0.568 ± 0.034 m). No other significant differences were found. Low to moderate TE (0.6–8.8%) were found for all variables, with SWC ranging 1.7–7.4%. The data provides evidence that the GPT can be used to measure kinetic and kinematic outputs, however, care should be taken when monitoring deadlift performance.
Article
Full-text available
Given two consecutive frames, video interpolation aims at generating intermediate frame(s) to form both spatially and temporally coherent video sequences. While most existing methods focus on single-frame interpolation, we propose an end-to-end convolutional neural network for variable-length multi-frame video interpolation, where the motion interpretation and occlusion reasoning are jointly modeled. We start by computing bi-directional optical flow between the input images using a U-Net architecture. These flows are then linearly combined at each time step to approximate the intermediate bi-directional optical flows. These approximate flows, however, only work well in locally smooth regions and produce artifacts around motion boundaries. To address this shortcoming, we employ another U-Net to refine the approximated flow and also predict soft visibility maps. Finally, the two input images are warped and linearly fused to form each intermediate frame. By applying the visibility maps to the warped images before fusion, we exclude the contribution of occluded pixels to the interpolated intermediate frame to avoid artifacts. Since none of our learned network parameters are time-dependent, our approach is able to produce as many intermediate frames as needed. We use 1,132 video clips with 240-fps, containing 300K individual video frames, to train our network. Experimental results on several datasets, predicting different numbers of interpolated frames, demonstrate that our approach performs consistently better than existing methods.
Article
Full-text available
Purpose: To describe the acute and delayed time course of recovery following resistance training (RT) protocols differing in the number of repetitions (R) performed in each set (S) out of the maximum possible number (P). Methods: Ten resistance-trained men undertook three RT protocols [S × R(P)]: (1) 3 × 5(10), (2) 6 × 5(10), and (3) 3 × 10(10) in the bench press (BP) and full squat (SQ) exercises. Selected mechanical and biochemical variables were assessed at seven time points (from - 12 h to + 72 h post-exercise). Countermovement jump height (CMJ) and movement velocity against the load that elicited a 1 m s(-1) mean propulsive velocity (V1) and 75% 1RM in the BP and SQ were used as mechanical indicators of neuromuscular performance. Results: Training to muscle failure in each set [3 × 10(10)], even when compared to completing the same total exercise volume [6 × 5(10)], resulted in a significantly higher acute decline of CMJ and velocity against the V1 and 75% 1RM loads in both BP and SQ. In contrast, recovery from the 3 × 5(10) and 6 × 5(10) protocols was significantly faster between 24 and 48 h post-exercise compared to 3 × 10(10). Markers of acute (ammonia, growth hormone) and delayed (creatine kinase) fatigue showed a markedly different course of recovery between protocols, suggesting that training to failure slows down recovery up to 24-48 h post-exercise. Conclusions: RT leading to failure considerably increases the time needed for the recovery of neuromuscular function and metabolic and hormonal homeostasis. Avoiding failure would allow athletes to be in a better neuromuscular condition to undertake a new training session or competition in a shorter period of time.
Article
Full-text available
The purpose of this study was to analyze the validity, reliability, and accuracy of new wearable and smartphone-based technology for the measurement of barbell velocity in resistance training exercises. To do this, 10 highly trained powerlifters (age = 26.1 ± 3.9 years) performed 11 repetitions with loads ranging 50–100% of the 1-Repetition maximum in the bench-press, full-squat, and hip-thrust exercises while barbell velocity was simultaneously measured using a linear transducer (LT), two Beast wearable devices (one placed on the subjects' wrist –BW–, and the other one directly attached to the barbell –BB–) and the iOS PowerLift app. Results showed a high correlation between the LT and BW (r = 0.94–0.98, SEE = 0.04–0.07 m • s −1), BB (r = 0.97–0.98, SEE = 0.04–0.05 m • s −1), and the PowerLift app (r = 0.97–0.98, SEE = 0.03–0.05 m • s −1) for the measurement of barbell velocity in the three exercises. Paired samples T-test revealed systematic biases between the LT and BW, BB and the app in the hip-thrust, between the LT and BW in the full-squat and between the LT and BB in the bench-press exercise (p < 0.001). Moreover, the analysis of the linear regression on the Bland-Altman plots showed that the differences between the LT and BW (R 2 = 0.004–0.03), BB (R 2 = 0.007–0.01), and the app (R 2 = 0.001–0.03) were similar across the whole range of velocities analyzed. Finally, the reliability of the BW (ICC = 0.910–0.988), BB (ICC = 0.922–0.990), and the app (ICC = 0.928–0.989) for the measurement of the two repetitions performed with each load were almost the same than that observed with the LT (ICC = 0.937–0.990). Both the Beast wearable device and the PowerLift app were highly valid, reliable, and accurate for the measurement of barbell velocity in the bench-press, full-squat, and hip-thrust exercises. These results could have potential practical applications for strength and conditioning coaches who wish to measure barbell velocity during resistance training.
Article
Full-text available
Purpose: to analyze the relationship between movement velocity and relative load (%1RM) in the pull-up exercise (PU), and to determine the pattern of repetition velocity loss during a single set to failure in pulling one's own body mass. Methods: Fifty-two men (age = 26.5 ± 3.9 years, body mass = 74.3 ± 7.2 kg) performed a first evaluation (T1) consisting of an one-repetition maximum test (1RM), and a test of maximum number of repetitions to failure pulling one's own body mass (MNR) in the PU exercise. Thirty-nine subjects performed both tests on a second occasion (T2) following 12 weeks' training. Results: We observed a strong relationship between mean propulsive velocity (MPV) and %1RM (r = -.96). Mean velocity attained with 1RM load (V1RM) was 0.20 ± 0.05 m·s(-1) and it influenced the MPV attained with each %1RM. Although 1RM increased by 3.4% from T1 to T2, the relationship between MPV and %1RM, and V1RM remained stable. We also confirmed stability in the V1RM regardless of individual relative strength. We found a strong relationship between percentage of velocity loss and percentage of performed repetitions (R(2) = .88), which remained stable despite a 15% increase in MNR. Conclusions: Monitoring repetition velocity allows estimation of the %1RM used as soon as the first repetition with a given load is performed, and the number of repetitions remaining in reserve when a given percentage of velocity loss is achieved during a PU exercise set.
Article
Full-text available
The purpose of this study was to analyse the validity and reliability of a novel iPhone app (named: PowerLift) for the measurement of mean velocity on the bench-press exercise. Additionally, the accuracy of the estimation of the 1-Repetition maximum (1RM) using the load–velocity relationship was tested. To do this, 10 powerlifters (Mean (SD): age = 26.5 ± 6.5 years; bench press 1RM · kg−1 = 1.34 ± 0.25) completed an incremental test on the bench-press exercise with 5 different loads (75–100% 1RM), while the mean velocity of the barbell was registered using a linear transducer (LT) and Powerlift. Results showed a very high correlation between the LT and the app (r = 0.94, SEE = 0.028 m · s−1) for the measurement of mean velocity. Bland–Altman plots (R2 = 0.011) and intraclass correlation coefficient (ICC = 0.965) revealed a very high agreement between both devices. A systematic bias by which the app registered slightly higher values than the LT (P < 0.05; mean difference (SD) between instruments = 0.008 ± 0.03 m · s−1). Finally, actual and estimated 1RM using the app were highly correlated (r = 0.98, mean difference (SD) = 5.5 ± 9.6 kg, P < 0.05). The app was found to be highly valid and reliable in comparison with a LT. These findings could have valuable practical applications for strength and conditioning coaches who wish to measure barbell velocity in the bench-press exercise.
Article
This study analyzed whether the loss of repetition velocity during a resistance exercise set was a reliable indicator of the number of repetitions left in reserve. Following the assessment of one-repetition (1RM) strength and full load-velocity relationship, thirty men were divided into three groups according to their 1RM strength/body mass: novice, well-trained and highly-trained. On two separate occasions and in random order, subjects performed tests of maximal number of repetitions to failure against loads of 65%, 75% and 85% 1RM in four exercises: bench press, full squat, prone bench pull and shoulder press. For each exercise, and regardless of the load being used, the absolute velocities associated to stopping a set before failure, leaving a certain number of repetitions (2, 4, 6 or 8) in reserve, were very similar and showed a high reliability (CV 4.4-8.0%). No significant differences in these stopping velocities were observed for any resistance training exercise analyzed between the novice, well-trained and highly-trained groups. These results indicate that by monitoring repetition velocity one can estimate with high accuracy the proximity of muscle failure and, therefore, to more objectively quantify the level of effort and fatigue being incurred during resistance training. This method emerges as a substantial improvement over the use of perceived exertion to gauge the number of repetitions left in reserve.
Article
The use of bar velocity to estimate relative load in the back squat exercise was examined. Eighty strength-trained men performed a progressive loading test to determine their one-repetition maximum (1RM) and load-velocity relationship. Mean (MV), mean propulsive (MPV) and peak (PV) velocity measures of the concentric phase were analyzed. Both MV and MPV showed a very close relationship to %1RM (R2 = 0.96), whereas a weaker association (R2 = 0.79) and larger SEE (0.14 vs. 0.06 m•s-1) was found for PV. Prediction equations to estimate load from velocity were obtained. When dividing the sample into three groups of different relative strength (1RM/body mass), no differences were found between groups for the MPV attained against each %1RM. MV attained with the 1RM was 0.32 ± 0.03 m•s-1. The propulsive phase accounted for 82% of concentric duration at 40% 1RM, and progressively increased until reaching 100% at 1RM. Provided that repetitions are performed at maximal intended velocity, a good estimation of load (%1RM) can be obtained from mean velocity as soon as the first repetition is completed. This finding provides an alternative to the often demanding, time-consuming and interfering 1RM or nRM tests and allows to implement a velocity-based resistance training approach.
Article
To determine the effect of circadian rhythm on neuromuscular responses and kinematics related to physical tennis performance, after a standardised warm-up, 13 highly competitive male tennis players were tested twice for serve velocity/accuracy (SVA), countermovement vertical jump (CMJ), isometric handgrip strength (IS), agility T-test (AGIL) and a 10-m sprint (10-m RUN). In a randomised, counter-balance order, tennis players underwent the test battery twice, either in the morning (i.e., AM; 9:00 h) and in the afternoon (i.e., PM; 16:30 h). Paired t-tests were used to analyse differences due to time-of-day in performance variables. Comparison of morning versus afternoon testing revealed that SVA (168.5 ± 6.5 vs. 175.2 ± 6.1 km · h⁻¹; P = 0.003; effect size [ES] = 1.07), CMJ (32.2 ± 0.9 vs. 33.7 ± 1.1 cm; P = 0.018; ES = 1.46), AGIL (10.14 ± 0.1 vs. 9.91 ± 0.2 s; P = 0.007; ES = 1.23) and 10-m RUN time (1.74 ± 0.1 vs. 1.69 ± 0.1 s; P = 0.021; ES = 0.67) were significantly blunted during the morning testing. However, IS was not affected by time-of-day (P = 0.891). Thus, tennis performance may be reduced when competing in the morning in comparison to early evening. Therefore, coaches and tennis players should focus on schedule the SVA, power, speed and agility training sessions in the afternoon.