Vision AI is a powerful tool that is changing the way that we interact with the digital world, and how users can interact with the apps that we build. With Vision AI, we can build apps that are smarter and more intuitive, allowing users to perform once cumbersome or complicated tasks with just a few images.
Today, I’m going to walk through an example of how that could work in a mock nutrition tracking app that we are building. Using Vision AI, we’ll be able to add the ability to automatically identify foods on your user’s plates from just an image, and then plug it into our existing nutritional database. This image recognition feature can greatly simplify the process of logging food intake for your users.
It’s traditionally been quite challenging to use Vision AI because of how to get the data to train computer vision models. However, with Dragoneye’s latest Recognize Anything models, it’s now possible to create a custom model that can recognize anything we want in just a few minutes.
In this article, we’ll walk you through the process and you'll be able to see just how simple it is to enhance your app’s capabilities with Vision AI.
To start, we’re going to have to build the model. Normally, this would mean collecting thousands to millions of images of food items and then labeling them one by one - something that could take months. However, with Dragoneye’s Recognize Anything models, we can skip all of that!
To start, let’s head over to the Dragoneye dashboard and the Recognize Anything page in the sidebar.
Here, we’re able to see any models that we’ve created already. Looks like we don’t have any models at the moment, but we can change that! Let’s click the Create button in the top right corner.
For our nutrition tracking app, we need the model to recognize various types of food, so we’ll begin by adding a list of the food item categories we want to identify.
Let’s say that we want to recognize the following categories:
We can input each into the model builder interface. After each category input, the builder will show you a preview of example images from that category. For example, “spaghetti bolognese” shows:
Once we’ve put in all of the categories, we can now hit Deploy. And, within just a few short seconds,, our custom Vision AI model is live and ready to be integrated into the app.
Now comes the fun part—testing our new model. To do this, we can use the interactive demo tool.
Let’s try it out with a few example images:
We can also see what the API request and response, which will be helpful when we go to integrate this into our app!
Hopefully this guide demonstrates how easy it is to build your own models using Dragoneye’s Recognize Anything platform. Simply by specifying the types of food we want to recognize, we can create a computer vision model that can quickly and accurately recognize the food in our user’s meals, making it a lot easier to keep track.
You can get started on your own custom model in the Dragoneye dashboard too! Happy building!
