Case Study - Food Detection & Nutrition Prediction
ML models for detecting foods and predicting calories and nutrients from images.
- Client
- OrthoFoodie
- Year
- Service
- Machine Learning
Overview
We're helping OrthoFoodie build ML models that detect foods and predict calories, nutrients, and other nutritional information from single images.
Our current model achieves 36% PMAE on nutrition5k — a strong result for single image monocular estimation. On warm GPUs with just an A100-40GB, inference takes approximately 0.5 seconds.
More on how we're building this soon.
What we did
- Computer Vision
- Machine Learning
- Nutrition Prediction
- PMAE on nutrition5k
- 36%
- Inference time (A100)
- 0.5s