Image Classification with AutoTrain

So you’ve heard all about the cool things that are happening in the machine learning world, and you want to join in. There’s just one problem – you don’t know how to code! 😱 Or maybe you’re a seasoned software engineer who wants to add some ML to your side-project, but you don’t have the time to pick up a whole new tech stack! For many people, the technical barriers to picking up machine learning feel insurmountable. That’s why Hugging Face created AutoTrain, and with the latest feature we’ve just added, we’re making “no-code” machine learning better than ever. Best of all, you can create your first project for ✨ free! ✨

Hugging Face AutoTrain lets you train models with zero configuration needed. Just choose your task (translation? how about question answering?), upload your data, and let Hugging Face do the rest of the work! By letting AutoTrain experiment with number of different models, there's even a good chance that you'll end up with a model that performs better than a model that's been hand-trained by an engineer 🤯 We’ve been expanding the number of tasks that we support, and we’re proud to announce that you can now use AutoTrain for Computer Vision! Image Classification is the latest task we’ve added, with more on the way. But what does this mean for you?

Image Classification models learn to categorize images, meaning that you can train one of these models to label any image. Do you want a model that can recognize signatures? Distinguish bird species? Identify plant diseases? As long as you can find an appropriate dataset, an image classification model has you covered.

How can you train your own image classifier?

Once the data has been added, simply choose the number of model candidates that you’d like AutoModel to try out, review the expected training cost (training with 5 candidate models and less than 500 images is free 🤩), and start training!