Dieto is pioneering a new approach in mobile nutrition tracking by combining on-device computer vision and text recognition to deliver instant, offline calorie estimates, marking a shift towards privacy-focused, real-time health apps.
Dieto is being positioned as a fully offline Android calorie-tracking app that tries to remove the friction that still defines much of nutrition logging. Instead of asking users to type in every ingredient or wait for a cloud request to resolve, the project combines on-device computer vision with text recognition to identify food, estimate portions and return nutrition guidance in real time. The developer behind the app says the aim is to deliver a premium scanner that can work directly on mobile hardware without depending on constant internet access.
At the heart of the system are two separate machine-learning pipelines. According to the project description, one uses YOLOv8 for object detection, while Google ML Kit handles optical character recognition, allowing the app to read packaging and other text from camera input. That dual-engine approach is intended to improve accuracy across a meal plate, where visual identification alone can be unreliable. The app then uses those detections to estimate serving sizes and scale nutritional values before presenting diet advice tailored to the user.
The broader market suggests there is real appetite for this kind of automation. Apps such as Caloi AI, Yazio, AI Calorie Counter & Deficit, Dietor and CalFuel all promote AI-assisted meal scanning, calorie estimates, macro tracking and faster food logging, while the App Store listing for Dieto itself says the app also personalises plans through lifestyle questions and photo-based meal analysis. What distinguishes Dieto in the developer’s account is the emphasis on running the core experience locally, rather than treating AI food recognition as a cloud service.
That design choice reflects a wider shift in mobile computing. As smartphone processors and neural engines have become more capable, developers are increasingly exploring how far vision models can be pushed on-device, especially for tasks where speed, privacy and offline availability matter. Dieto’s architecture is framed as an answer to the shortcomings of older calorie apps, which often still depend on manual entry or remote databases that can slow the user experience and make tracking feel more like admin than support.
The project is therefore less a single app announcement than a demonstration of how consumer AI is changing the rules of mobile health software. By combining detection, text extraction and nutrition estimation in one workflow, Dieto aims to make calorie tracking feel immediate and automatic. Whether that promise translates into widespread adoption will depend on how well it performs outside the lab, but the technical ambition is clear: to make a pocket-sized diet scanner that behaves more like a live assistant than a conventional logging tool.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
