AI Sports Coaching Glossary

A computer vision technique that identifies and tracks the position of key anatomical landmarks — typically 17 body joints including shoulders, elbows, hips, and knees — across video frames. By mapping these landmarks in real time, AI systems calculate joint angles, movement velocity, and postural alignment. Applications like SportsReflector use pose estimation to score athletic technique on a 0–100 scale without requiring manual annotation. [1] Cao et al., IEEE TPAMI, 2019 (arXiv:1812.08008)

The process of locating specific anatomical landmarks — joints such as shoulders, elbows, wrists, hips, knees, and ankles — within individual video frames. Deep learning models trained on large annotated datasets output keypoint coordinates that downstream systems use to construct skeletal models. SportsReflector detects 25+ keypoints per frame to enable real-time biomechanical scoring across 20+ sports.

A simplified representation of the human body as joints (nodes) connected by limb segments (edges), derived from keypoint detection. The skeletal model is overlaid on video footage to visualise body position and serves as the input for calculating joint angles, segment velocities, and movement timing. SportsReflector renders skeletal models in real time during AR-guided training sessions.

An approach to athletic instruction in which machine learning models analyse video footage to detect body position, measure movement parameters, and deliver technique feedback — replacing or supplementing human observation. Computer vision coaching systems like SportsReflector process each video frame independently, enabling objective, consistent feedback across every repetition without a coach present.

A technique that reconstructs three-dimensional joint positions from video, capturing depth information unavailable in standard 2D analysis. Traditional 3D motion analysis required multi-camera labs with reflective markers; modern AI approaches estimate 3D positions from a single smartphone camera using deep learning models trained on large motion capture datasets.

The review of video footage one frame at a time to examine body position at specific moments during a movement. At 240 frames per second, each frame represents approximately 4 milliseconds — sufficient to capture the release point in a basketball shot or the moment of maximum knee flexion in a squat. SportsReflector supports frame-by-frame scrubbing for all recorded sessions.

The application of mechanical principles to the study of human movement, measuring joint angles, angular velocities, force vectors, and timing relationships between body segments. AI-powered biomechanical analysis automates measurements that previously required motion capture laboratories. SportsReflector delivers biomechanical analysis from a standard smartphone camera within seconds of recording. [2] Knudson, Fundamentals of Biomechanics, 2nd ed., Springer, 2013.

The sequence of body segments that transfer force from one to the next during an athletic movement. In a tennis serve, the chain runs from feet through legs, hips, trunk, shoulder, elbow, and wrist to the racquet. Breakdowns — such as early shoulder rotation or insufficient hip drive — reduce power and increase injury risk. SportsReflector identifies kinetic chain faults by analysing segment timing. [3] Kibler et al., Sports Medicine, 2006 (DOI:10.2165/00007256-200636020-00001)

The angle formed between two adjacent body segments at a joint, measured in degrees. Optimal joint angles vary by sport and movement phase — for example, 90° elbow flexion at the top of a basketball free throw, or 70–90° knee flexion at the bottom of a squat. AI coaching apps calculate joint angles from keypoint positions and compare them against sport-specific reference ranges.

The order and timing in which body segments reach their peak rotational velocity during a movement. In an efficient golf swing, the sequence progresses from pelvis to thorax to lead arm to club, with each segment decelerating to transfer energy to the next. Disruptions — such as the shoulders peaking before the hips — reduce power and consistency. [8] Putnam, J Biomech, 1993 (DOI:10.1016/0021-9290(93)90084-R)

The rotational lag between hips and shoulders during rotational athletic movements such as golf swings, baseball swings, and tennis groundstrokes. Effective separation allows the hips to initiate the downswing while the shoulders remain coiled, creating a stretch-shortening cycle that amplifies power output. AI analysis measures this separation in degrees and milliseconds.

The angular difference between hip rotation and shoulder rotation at the top of the backswing in golf, typically 45–55 degrees in professional players. A larger X-factor creates greater elastic energy in the trunk muscles, released during the downswing to generate clubhead speed. AI golf analysis measures X-factor by calculating hip and shoulder rotation angles at the transition point.

An inward deviation of the knee joint during weight-bearing movements such as squats, lunges, and landings, associated with elevated ACL injury risk. It is commonly caused by weak hip abductors, limited ankle dorsiflexion, or poor motor control. AI coaching apps like SportsReflector detect valgus collapse by measuring medial knee deviation relative to the hip and ankle keypoints.

A comparison of movement quality and joint angles between the left and right sides of the body. Asymmetries greater than 10–15% between sides are associated with elevated injury risk and compensatory movement patterns. SportsReflector's symmetry analysis flags significant left-right imbalances and tracks them over time to monitor whether training is correcting or worsening the asymmetry. [4] Hewit et al., J Strength Cond Res, 2012 (DOI:10.1519/JSC.0b013e318231a61a)

A measurable difference in movement quality, joint angles, or force production between corresponding left and right body segments during athletic activity. Asymmetries above 15% are clinically associated with overuse injury and performance loss. AI platforms like SportsReflector quantify movement asymmetry across every repetition and flag trends that warrant corrective intervention.

The extent of movement available at a joint, measured in degrees. Adequate range of motion is a prerequisite for correct technique in most sports — insufficient hip mobility limits squat depth, restricted shoulder mobility affects overhead pressing form, and limited thoracic rotation reduces golf swing turn. AI coaching apps measure range of motion during movements and flag restrictions that may indicate mobility limitations.

A visualisation that estimates which muscle groups are most active during a given movement, derived from joint angles, segment positions, and movement velocity. While not a direct EMG measurement, AI-based muscle activation mapping approximates primary and secondary muscle engagement. SportsReflector uses this to help athletes understand which muscles are driving — or failing to drive — their movements.

A colloquial term for posterior pelvic tilt at the bottom of a squat, where the pelvis tucks under the spine causing lumbar flexion. Butt wink increases compressive load on the lumbar discs and is commonly caused by limited hip flexion mobility or ankle dorsiflexion restriction. AI coaching apps detect it by tracking pelvic angle relative to the spine across video frames.

A numerical rating (0–100) that quantifies the quality of an athletic movement by comparing measured biomechanical parameters — joint angles, timing, symmetry, range of motion — against sport-specific reference models. SportsReflector calculates a form score for every recorded repetition; scores above 85 indicate good technique, while scores below 60 typically signal a significant mechanical issue.

An automated process in which machine learning models analyse video footage to measure movement parameters and assign a numerical quality score without human annotation. AI biomechanical scoring systems like SportsReflector evaluate joint angles, timing, symmetry, and range of motion simultaneously, producing objective scores that are consistent across every session and repeatable across athletes.

A composite metric that aggregates multiple biomechanical measurements into a single 0–100 rating for a specific athletic movement. Each sport has its own scoring model with differently weighted parameters — a basketball free throw weights elbow alignment, release angle, and follow-through, while a golf swing weights spine angle, hip rotation, and weight transfer.

Technique corrections generated by an AI system without human coach input, delivered immediately after or during a movement. Automated form feedback systems analyse video frames, identify deviations from reference biomechanics, and output prioritised corrections. SportsReflector delivers automated form feedback within seconds of recording, covering joint angles, timing, symmetry, and injury risk flags.

A biomechanical template representing optimal technique for a specific movement, derived from elite athlete analysis or established coaching principles. AI coaching apps compare measured parameters against the reference model to calculate form scores and identify deviations. Reference models are sport-specific and may vary by skill level, body type, and movement style.

A real-time visual layer superimposed on live camera footage displaying coaching cues, movement paths, joint angle targets, and form corrections. SportsReflector's AR overlay shows the ideal movement trajectory alongside the athlete's actual motion, enabling self-correction without a coach present. Overlays are rendered at the camera's native frame rate to maintain alignment with body position.

An augmented reality interface that projects sport-specific coaching cues — ideal joint positions, movement arcs, timing markers, and rep counts — directly onto live video of an athlete in motion. AR coaching overlays eliminate the need for a physical coach during practice by providing immediate visual guidance. SportsReflector delivers AR coaching overlays for 20+ sports and gym exercises.

Coaching corrections and form cues delivered during or immediately after a movement, before the athlete's next repetition. Real-time feedback accelerates motor learning by closing the gap between action and correction. SportsReflector delivers real-time feedback via on-screen cues during AR training and via instant score summaries after each recorded repetition.

An automated evaluation of movement patterns statistically associated with increased injury probability. Common flags include knee valgus during squats, excessive lumbar flexion during deadlifts, and early trunk rotation during throwing motions. AI injury risk assessment does not diagnose injuries but identifies movement patterns warranting attention from a medical professional. [5] Hewett et al., Am J Sports Med, 2005 (DOI:10.1177/0363546504269591)

An AI analysis feature that identifies when an athlete's form begins to deteriorate due to accumulated fatigue during a training session. SportsReflector's fatigue detection tracks form score trends across multiple repetitions and alerts the athlete when technique degradation reaches a threshold associated with elevated injury risk.

The simultaneous or sequential analysis of a movement from multiple camera angles — typically face-on and down-the-line for golf, or front and side views for gym exercises. Multi-angle analysis provides a more complete picture of movement quality, as some faults are only visible from specific angles. SportsReflector supports multi-angle video upload and combined scoring.

An AI technique that automatically counts exercise repetitions by detecting the cyclical pattern of joint positions across video frames, without requiring manual input or wearable sensors. Computer vision rep counting systems identify the start and end of each repetition by tracking key joints through their full range of motion. SportsReflector counts reps automatically during AR-guided gym sessions.

The process by which the nervous system acquires, refines, and stores movement patterns through practice. Effective motor learning requires accurate feedback delivered at the right time and frequency. AI coaching accelerates motor learning by providing objective, consistent feedback on every repetition, eliminating the subjective variability of human observation. [6] Schmidt & Lee, Motor Control and Learning, 6th ed., 2019. [7] Wulf & Shea, Psychol Bull, 2002 (DOI:10.1037/0033-2909.128.3.369)

An AI coaching approach in which a single platform analyses technique across multiple sports and fitness disciplines using shared computer vision infrastructure with sport-specific scoring models. Multi-sport AI coaching eliminates the need for separate apps per activity. SportsReflector covers 20+ sports — including basketball, tennis, golf, boxing, and gym exercises — within a single application.

Software tools that enable athletes to analyse, score, and improve their own technique without access to a professional coach. Self-coached athlete technology typically combines video analysis, AI form scoring, corrective drill libraries, and progress tracking. SportsReflector is designed specifically for self-coached athletes, delivering professional-grade biomechanical feedback from a standard smartphone.