Being able to capture relevant information about elite athletes’ movement “in the wild” is challenging, especially because reference marker-based approaches hinder natural movement and are highly sensitive to environmental conditions. We propose Pose2Sim, a markerless kinematics workflow that uses OpenPose 2D pose detections from multiple views as inputs, identifies the person of interest, robustly triangulates joint coordinates from calibrated cameras, and feeds those to a 3D inverse kinematic full-body OpenSim model in order to compute biomechanically congruent joint angles. We assessed the robustness of this workflow when facing simulated challenging conditions: (Im) degrades image quality (11-pixel Gaussian blur and 0.5 gamma compression); (4c) uses few cameras (4 vs. 8); and (Cal) introduces calibration errors (1 cm vs. perfect calibration). Three physical activities were investigated: walking, running, and cycling. When averaged over all joint angles, stride-to-stride standard deviations lay between 1.7° and 3.2° for all conditions and tasks, and mean absolute errors (compared to the reference condition—Ref) ranged between 0.35° and 1.6°. For walking, errors in the sagittal plane were: 1.5°, 0.90°, 0.19° for (Im), (4c), and (Cal), respectively. In conclusion, Pose2Sim provides a simple and robust markerless kinematics analysis from a network of calibrated cameras.

Two-dimensional deep-learning pose estimation algorithms can suffer from biases in joint pose localizations, which are reflected in triangulated coordinates, and then in 3D joint angle estimation. Pose2Sim, our robust markerless kinematics workflow, comes with a physically consistent OpenSim skeletal model, meant to mitigate these errors. Its accuracy was concurrently validated against a reference marker-based method. Lower-limb joint angles were estimated over three tasks (walking, running, and cycling) performed multiple times by one participant. When averaged over all joint angles, the coefficient of multiple correlation (CMC) remained above 0.9 in the sagittal plane, except for the hip in running, which suffered from a systematic 15° offset (CMC = 0.65), and for the ankle in cycling, which was partially occluded (CMC = 0.75). When averaged over all joint angles and all degrees of freedom, mean errors were 3.0°, 4.1°, and 4.0°, in walking, running, and cycling, respectively; and range of motion errors were 2.7°, 2.3°, and 4.3°, respectively. Given the magnitude of error traditionally reported in joint angles computed from a marker-based optoelectronic system, Pose2Sim is deemed accurate enough for the analysis of lower-body kinematics in walking, cycling, and running.

Speed climbing involves an optimization of the velocity of the ascent and the trajectory path during performance. Consequently, any amount of energy spent in the two other directions than vertical, namely the lateral direction and the direction perpendicular to the wall plane, is a potential loss of performance. To assess this principle, we present a study on 3D motion analysis and its 3D visualization for a subject during a speed climbing performance. The fundamentals of geometrical measurement in 3D require to integrate multiple 2D cues, at least two, in order to extract 3D information. First results with two drones following an athlete's ascent show that a 3D velocity profile can be provided from the tracking of a marker on the harness, pointing critical phases in the ascent where the vertical speed is not dominant any more. We further investigate 3D motion of full body using markerless video-based tracking. Our approach is based on a full body 3D avatar model of the climber, represented as a 3D mesh. This model and its deformation are learned in a laboratory studio. The learning needs to be done only once. Result is a manifold embedding of the 3D mesh and its deformations, which can be used afterwards to perform registration onto video of performance of speed climbing. The results of the tracking is an inference of the 3D mesh aligned onto videos of speed climbing performance. From this 3D mesh, we deduce an estimation of the center of mass (COM). We show that this estimation from 3D mesh differs from the usual approximation of the COM as a marker on the harness. In particular, the 3D mesh COM takes into account the whole body movement such as the influence of the limbs which is not detected by a marker on the harness.

Inertial sensors are increasingly used in rodent research, in particular for estimating head orientation relative to gravity, or head tilt. Despite this growing interest, the accuracy of tilt estimates computed from rodent head inertial data has never been assessed. Using readily available inertial measurement units mounted onto the head of freely moving rats, we benchmarked a set of tilt estimation methods against concurrent 3D optical motion capture. We show that, while low-pass filtered head acceleration signals only provided reliable tilt estimates in static conditions, sensor calibration combined with an appropriate choice of orientation filter and parameters could yield average tilt estimation errors below $1.5$${}^{\circ }$ during movement. We then illustrate an application of inertial head tilt measurements in a preclinical rat model of unilateral vestibular lesion and propose a set of metrics describing the severity of associated postural and motor symptoms and the time course of recovery. We conclude that headborne inertial sensors are an attractive tool for quantitative rodent behavioral analysis in general and for the study of vestibulo-postural functions in particular.