Why limit yourself to your small screen on the phone? Run Pocket AutoML, an app by Evgeniy Mamchenko, best experienced on your PC or Mac with BlueStacks, the world’s #1 Android emulator.
Pocket AutoML feels like a tiny lab for teaching a device to recognize things, without needing to code. Open it, point it at a folder of photos, make a couple of labels, and hit train. The app builds an image classifier in seconds for small sets, then lets the user test it right away. It keeps the techy stuff tucked away, so even though it uses a real convolutional neural network under the hood, the flow is simple. On a PC with BlueStacks, the larger screen makes sorting and labeling images easier, and using a mouse for quick taps helps a lot. Results are surprisingly decent with only a few photos per label, which is that few-shot learning idea, and that makes the whole thing feel fast and practical.
Privacy is a big part of it. Training and predictions happen on the device, no internet needed, no image uploads lurking in the background. People who are deeper into computer vision can still get mileage here, because it is good for quick transfer learning tests without writing a single line of code. The model export to TensorFlow Lite is there too, so someone can take what they trained and plug it into a simple Android app following the built-in guidance. It is not flashy and it does not try to be. It is a straightforward tool that lets a person try an idea, see if it works, and move on. That is the appeal.
Big screen. Bigger performance. Use BlueStacks on your PC or Mac to run your favorite apps.
