-->
When you take a picture with Cal AI, many things happen. First the depth sensor on your phone will figure out the volume of your food. Using that information, our AI model which was trained on thousands of food images will seperate your meal into all the different parts and their proportions. Finally, our multimodal AI model will take everything in and finalize the results: figuring out the calories, protein, carbs, and fat in your meal.
Cal AI's image scanning feature is generally 90% accurate. That said, if you scan a food with hidden ingredients, like a smoothie or a soup, it will have a harder time. For such foods, you are better off using our describe feature which lets you type exactly what you ate, with any relevant information. You also could use our barcode scan feature or the food label scan feature. In case a result comes back inaccurate, you can always click the 'fix results' button to describe what was wrong so that our AI can fix it. This also helps us get better :)