AI-Powered Body Fat Estimation, Made Simple.
BodyFat is a mobile-first app that uses computer vision and on-device AI to estimate body fat percentage directly from photos — fast, private, and accessible. Designed for everyday users, fitness enthusiasts, and health-conscious individuals, BodyFat removes the need for expensive hardware, calipers, or clinic visits by turning a smartphone camera into a smart body composition tool. Our scope focused on building a reliable, intuitive experience that balances accuracy with ease of use. We implemented the full mobile pipeline — from guided image capture and pose normalization to AI-based body analysis and clear result visualization — all optimized to run smoothly on consumer devices. Special attention was given to UX clarity, privacy-first processing, and performance across a wide range of phones. Estimate body fat using photos — no equipment required Guided capture flow for consistent, repeatable results On-device AI processing with a privacy-first approach Clean, simple results tailored for non-technical users BodyFat delivers a modern, approachable way to understand body composition — right from your phone.
Industries:
AI, Computer Vision, Fitness
Country:
USA
Services:
AI, Computer Vision
Team involved:
React Native Developer
Duration:
5 month
Technologies

Challenges
Solutions
The project required a single pose detection solution that behaves consistently across both platforms, with identical joint tracking and skeletal structure.
Implemented Google’s MediaPipe Pose model as the core detection layer, enabling reliable, cross-platform skeletal tracking on both iOS and Android.
The model outputs only skeletal landmarks. Higher-level data such as joint angles, limb relationships, and pose correctness had to be derived manually.
Built a mathematical processing layer on top of MediaPipe landmarks to compute joint angles, limb orientation, and movement dynamics for accurate pose analysis.
Continuous pose detection and calculations risked frame drops and UI lag, especially on mid-range devices.
Moved pose inference and calculations off the main thread, ensuring smooth UI updates and stable real-time performance during continuous tracking.
Results
Great Work!
Liam WilsonFounder @ BodyFat
Selected Work
Apps used by real people — built from scratch by our team.

CloPinz Golf
Closest-To-The-Pin Contests-Reinvented
View Case

ShotSense Golf
Play smarter and score lower with advanced golf intelligence for every shot
View Case

Vera
Elevate the mall experience with AI and AR-powered intelligent building management.
View Case

Evooia
Patient-focused mobile app with a powerful web platform for doctors.
View Case

