The countermovement jump (CMJ) assessment is widely employed for monitoring sports performance, traditionally relying on heavy and expensive force plates to extract performance variables like jump height and peak force. Inertial measurement unit (IMU)-based approaches and mobile applications have been developed to analyse CMJ performance with cost-effective devices, but they still require technical expertise and manual annotations during operation. We developed a new camera-based pipeline that can measure CMJ performance automatically by utilising computer vision techniques and biomechanical approaches from video captured by a single uncalibrated camera. Human segmentation and pose estimation techniques are used to understand the movement of the centre of mass and take-off and landing times. Combined with the biomechanical principles of object parabolic motion and inverse dynamics, the force–time data can be estimated for extracting CMJ performance variables. We recruited 77 elite athletes (29 females; height: 170.0 ± 9.0 cm; mass: 72.2 ± 17.7 kg) to evaluate the developed method against a commercial force platform. The developed method enables fully automatic CMJ analysis for both force–time data and performance variables from video captured by a camera without calibration. The results showed superior correlations (R > 0.7) and high reliability (%CV < 10 %) for most CMJ variables compared to the IMU-based approach. This approach automates CMJ analysis, offering more variables than existing mobile apps while reducing the technical demands of IMU-based methods. It streamlines assessment, making it ideal for large-scale cohort studies. Grounded in biomechanics, it enhances sports and health monitoring, enabling data-driven optimisation of human performance.

Computer vision

Machine learning

Biomechanics

Analysis

Testing

© 2025 The Authors. Published by Elsevier Ltd.