Body Action Counting SDK
The Body Action Counting SDK uses a camera to capture human movements, identify human keypoints in real time, and automatically count action repetitions. It supports 15 fitness actions, including jump rope, squats, jumping jacks, sit-ups, push-ups, planks, and glute bridges.
As of May 12, 2026, the Alibaba Cloud Vision Intelligence Open Platform will no longer issue new license authorizations for its offline SDK products. Customers who have previously purchased these services can continue to use them. However, no new license authorizations will be issued after this date.
Advantages
High flexibility: Outputs data for each of the 15 human keypoints individually. You can also combine multiple keypoints to count repetitions of a fitness action, with an average accuracy of over 95%.
High adaptability: Its template-based configuration allows you to quickly adapt the SDK for new actions. It supports real-time negative feedback to help users correct their form. The SDK performs reliably in various scenarios, lighting conditions, and backgrounds.
Efficient performance: The algorithm is easy to use and integrates quickly into your applications. It provides high frame-to-frame stability while using minimal CPU and memory resources. On entry-level devices, it runs at over 20 frames per second (fps).
Use cases
Fitness apps: Use the device camera to capture and count fitness actions in real time. It acts as a training assistant, allowing users to review their form and accuracy during playback.
Smart fitness mirrors: Capture and count fitness actions in real time using the mirror's built-in camera. Users can improve their workout quality by making immediate adjustments based on negative feedback and their own reflection.
Prerequisites
Before using the SDK, you must activate and authorize it. For more information, see Activate and authorize offline SDKs.
Pricing
For information about the pricing of the Body Action Counting SDK, see Offline SDK pricing.
Supported devices
Android and iOS
Constraints and restrictions
Android: Android 5.0 or later. The device must include a camera module and a gyroscope module.
iOS: iOS 10 or later. Supports iPhone 5s and later models. The device must include a camera module.
Recommended camera resolution: The native camera resolution should be between 640x480 and 1280x720 pixels. A higher resolution reduces the speed of the SDK.
To use the SDK on iPhone 5s or earlier models, or on low-end Android devices, contact us.
The SDK supports only one person in the frame at a time. Their full body must be visible and occupy at least one-quarter of the screen.
Input data resolution: Between 640x480 and 1280x720 pixels.
The user must face the camera.
Integration
This SDK does not transmit any data to the SDK provider. Therefore, if you use this SDK to collect personal information, you are considered the data processor. The SDK provider is not a data processor or a sub-processor. You must fully inform users and obtain their authorization in accordance with relevant laws and regulations. When informing users, ensure that the SDK provider is not identified as a data processor or a sub-processor.
Android: See the Android integration development guide.
iOS: See the iOS integration development guide.
Appendix
Figure 1. Human keypoints diagram
The SDK supports the following 15 fitness actions.
Portrait mode actions: Jump rope, squats, jumping jacks, squat jacks, seal jacks, star jumps, diamond stretch, arm circles, and waist circles.
Landscape mode actions: Sit-ups, crunches, push-ups, knee push-ups, and glute bridges.
Open source models
To experience and download more free open source models, visit the ModelScope community: HRNet for 2D human keypoints and 2D full-body keypoint detection for general scenarios.