Dark Mode
Exercise Joint Angle Dataset
Public Health & Epidemiology
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This dataset focuses on analyzing human body movements during common exercises by capturing and processing angles of key body joints. It uses video data to extract frame-by-frame angles for exercises such as push-ups, jumping jacks, pull-ups, squats, and Russian twists. The methodology involves MediaPipe for pose estimation to detect body landmarks and YOLOv6 for object detection, enhancing the accuracy and robustness of the analysis. The primary purpose is to provide data that can be used to detect exercises and to facilitate the development of intelligent fitness assistants and advanced movement analysis tools.
Columns
- Side: Indicates the side from which the movement was captured. The most common value is "left".
- Shoulder_Angle: Represents the angle measured between the upper arm and the torso. Values typically range from 0 to 180 degrees, with a mean of 66.5.
- Elbow_Angle: Represents the angle measured between the upper arm and forearm. Values typically range from 0 to 180 degrees, with a mean of 114.
- Hip_Angle: Represents the angle measured between the torso and the upper leg. Values typically range from 0.01 to 180 degrees, with a mean of 137.
- Knee_Angle: Represents the angle measured between the upper leg and the lower leg. Values typically range from 0.12 to 180 degrees, with a mean of 143.
- Ankle_Angle: Represents the angle measured between the lower leg and the foot. Values typically range from 0.03 to 180 degrees, with a mean of 135.
- Shoulder_Ground_Angle: Represents the angle measured between the shoulder and the ground plane. Values typically range from -90 to 90 degrees, with a mean of 88.8.
- Elbow_Ground_Angle: Represents the angle measured between the elbow and the ground plane. Values typically range from -90 to 90 degrees, with a mean of 88.9.
- Hip_Ground_Angle: Represents the angle measured between the hip and the ground plane. Values typically range from -90 to 90 degrees, with a mean of 79.4.
- Knee_Ground_Angle: Represents the angle measured between the knee and the ground plane. Values typically range from -90 to 90 degrees, with a mean of 75.8.
- Ankle_Ground_Angle: Represents the angle measured between the ankle and the ground plane. Values typically range from -90 to 90 degrees, with a mean of 69.
- Label: Categorises the specific exercise being performed. Examples include 'Push Ups' (31% of records) and 'Pull ups' (21% of records), with 5 unique exercise labels in total.
Distribution
The dataset is provided in a CSV format, named
exercise_angles.csv, with a file size of 4.17 MB. It consists of 12 distinct columns. Most columns contain 31,000 valid records, indicating a substantial volume of data for analysis. The structure allows for time-series analysis of joint movements during exercises.Usage
This dataset offers a variety of ideal applications. It can be utilised for form correction by comparing observed joint angles against standard benchmarks to provide feedback on technique. Users can also engage in performance tracking to monitor improvements in their exercise form over time. For machine learning, it serves as input for pose classification models to differentiate between correct and incorrect exercise execution, enabling the development of smart fitness assistants. Furthermore, it supports the creation of real-time feedback systems for live workout guidance, time-series analysis to identify movement trends, and pose optimisation models to suggest improvements in exercise form.
Coverage
The dataset focuses on the analysis of specific exercises: Push-ups (tracking shoulder, elbow, and hip angles), Jumping Jacks (tracking full-body motion including shoulder, elbow, hip, knee, and ankle angles), Pull-ups (tracking shoulder and elbow joint movements), Squats (analyzing hip, knee, and ankle angles for depth and posture), and Russian Twists (tracking core movement via shoulder and hip angles). Details regarding geographic, time range, or demographic scope are not specified in the provided sources.
License
CC0: Public Domain
Who Can Use It
This dataset is valuable for machine learning engineers developing models for pose classification and exercise detection. Fitness app developers can use it to build real-time feedback systems and smart workout assistants. Sports scientists may apply it for biomechanical research and performance analysis. Physical therapists could use it to track patient rehabilitation progress and improve exercise form. It is also suitable for researchers and data scientists interested in human movement analysis and exercise science.
Dataset Name Suggestions
- Exercise Joint Angle Dataset
- Human Exercise Kinematics
- Workout Movement Analysis
- Fitness Pose Estimation Angles
- Joint Dynamics for Exercise
Attributes
Original Data Source: Exercise Joint Angle Dataset
QUALITY