ANNIE.ML is a website that offers you a fun way to experience artificial intelligence, by enabling you to run your own AI projects directly in the web browser.
AnnieAI is a web-based AI project using TensorFlow by Google, which allows you to create your own AI playgrounds. The idea behind the project is to ease entry into the AI world for everyone by letting you create your own small AI applications.
Annie is your personal artificial friend that is excited to learn new things. With our playground, you can teach Annie about objects in your room or bring your friends and let them meet Annie.
Using the playground, you can add pictures with your webcam and train the AI in your web browser. By creating new classes, you can, e.g., add new objects you have lying around in your room. For example, an Apple and an ML learning Book.
For visualization, let's take the Apple and the ML Learning Book. The default amount of classes we have is two however it is up to you how many classes you want to create. For our example, we will work with the two classes already created.
Let's start with the Apple. Our website uses the webcam to take pictures of the objects you want to add, so please allow the website access. Now that your webcam picture appears, you can take the apple and hold it into the frame. Start taking pictures by pressing and holding the "Take Picture" button. You will see the number of images you've taken on the right. When you shake your hand, or some pictures are lower quality, don't be mad because that helps Annie to detect complex or lower quality pictures of your object in the future. Let's say that you took 100 photos of the apple, you may now change the object. Our second object is the ML Learning Book. Position it inside the webcam frame and hold the "Take Picture." button from the second class. For accuracy reasons, please make sure that you take roughly the same amount of pictures for every item. In our case, we should have around One hundred photos of each object.
Now let's test the AI. Take the item of your choice, and Annie now tells you how sure she is about what item she believes you are showing. By adding more evenly distributed pictures, the accuracy becomes better.
To deploy this project run
Clone Project:
git clone https://github.com/justTil/xu-web-ai-project-22.gitDownload Dependencies
npm iStart
npm run devCaution: When starting the project for the first time, an error concerning Firebase may appear. If that happens, reload the page and the error should be gone.
Client: React, Next.js, MUI
Server: Next.js, Vercel, Firebase