Download PDF
Apolo > Case Studies > Altis: Revolutionizing Home Fitness with AI Personal Trainer
Apolo Logo

Altis: Revolutionizing Home Fitness with AI Personal Trainer

Technology Category
  • Analytics & Modeling - Real Time Analytics
  • Cybersecurity & Privacy - Identity & Authentication Management
Applicable Industries
  • Education
  • Oil & Gas
Applicable Functions
  • Product Research & Development
Use Cases
  • Computer Vision
  • Real-Time Location System (RTLS)
Services
  • Cloud Planning, Design & Implementation Services
  • Training
The Challenge

Altis, an innovative consumer AI startup, aimed to revolutionize the home fitness industry by offering personalized fitness training using AI vision technology. The proposed AI system needed to accurately track and analyze movements, identify a wide range of exercises, including those using popular gym equipment, weights, machines, and detect errors in form. The system also needed to support multiple cameras and operate in real-time. To control costs and maximize asset utilization, Altis wanted to implement pipelines and MLOps on their existing on-premise GPU servers, while retaining the ability to scale globally on the cloud of their choice. The challenge was to develop a solution that was suitable for both hardware and cloud-based inference, and could scale globally without being tied to a single cloud provider.

About The Customer

Altis is a pioneering consumer AI startup with the ambitious goal of transforming the home fitness industry. In the post-pandemic, work-from-anywhere world, home fitness is booming, and Altis aims to capitalize on this trend by offering personalized fitness training in the comfort of the user's home. Altis' innovative product is a sleek soundbar-sized console that connects to any screen in the home. The console uses advanced AI and computer vision to provide personalized workouts and interactive coaching, helping users understand how their body is moving during exercise and improve form and movement performance.

The Solution

Neu.ro researchers developed a custom data pipeline that employs volumetric triangulation from one or more RGB cameras, combines 2D backbone data into volumetric aggregation of an intermediate 2D feature map, followed by refinement via 3D convolutions into a 3D heatmap and pose model. The system runs in real-time on Nvidia 3090 hardware and runs at 75 FPS while pose tracking, but has a startup mode of 30 FPS that runs until a person is detected. This volumetric model is able to estimate 3D human pose using any number of cameras, even using only 1 camera. In single-view setup, results are comparable with the current state of the art, while multiple sensors result in faster processing and fewer errors from potential occlusions.

Operational Impact
  • The implementation of the AI system has revolutionized the way Altis offers fitness training. The system's ability to accurately track and analyze movements, identify a wide range of exercises, and detect errors in form has significantly improved the quality of fitness instruction provided. The use of multiple cameras has enhanced the system's accuracy and speed, resulting in faster processing and fewer errors from potential occlusions. The system's ability to run on both hardware and cloud-based inference has provided Altis with the flexibility to scale globally without being tied to a single cloud provider. This has not only maximized asset utilization but also controlled costs.

Quantitative Benefit
  • The resulting model achieved a median error of less than 3cm per keypoint.

  • The system runs at 75 FPS while pose tracking.

  • In single-view setup, results are comparable with the current state of the art.

Related Case Studies.

Contact us

Let's talk!

* Required
* Required
* Required
* Invalid email address
By submitting this form, you agree that IoT ONE may contact you with insights and marketing messaging.
No thanks, I don't want to receive any marketing emails from IoT ONE.
Submit

Thank you for your message!
We will contact you soon.