Read more about this page and how it came to be in a blog post describing it in more detail.

Use the inputs below to train the model based on what is visible in the camera input/video area. Take several pictures of each label you would like to classify, but try to keep the numbers roughly equal. Move the object around and take several pictures at different angles/distances (depending on what you're trying to classify). Once you have the number of sample images you would like, click the Train button. Once training is complete, you can click the Start guessing button to begin to use the model to guess what is currently in the camera input.

Label: ...
Confidence: ...

You can save this model for use later (you will download both a model.json file and model.weights.json file). It can be uploaded using the form below to allow you to start guessing right away without training.

Upload both the model.json and model.weights.json files that were downloaded using the Save model button above

🏗 Built with ml5.js and p5.js 🏗