
Android app monetization tips, unexpectedly got some traction
I Built an Offline Bug-Bite Classifier for Android
Some time ago I made a small Android project to experiment with image classification.
The idea was simple: take a photo of a suspected bug bite, run a small model directly on the phone, and return the most likely type.
It became Bugbite Identifier, and somehow it has now passed 10K downloads on Google Play, even though I did basically no marketing.
The classifier runs fully offline using TensorFlow Lite. No image needs to be sent to an API, which also means there is basically no inference cost for me.
It currently recognizes mosquito, bed bug, flea, tick, spider, chigger, ant, and “not a bug bite”.
The hardest part was not really MobileNet or Android integration, but the dataset. Bug bites can look very different depending on skin, lighting, camera, how old the bite is, etc. A model can look good on a validation set and still behave quite differently on random real photos.
Now that the app has some users, I'm wondering if there is a small business hidden in this side project.
It currently has ads, but I'm not sure what makes the most sense next. Maybe a small one-time payment to remove ads, or some extra paid features. A subscription feels a bit strange since everything runs locally and there isn't really a recurring cost.
If you had a niche Android utility with 10K+ mostly organic downloads, how would you monetize it?
The app is bugbite identifier Google Play, free for now.
If anyone is interested, I can also write something about the MobileNet/TFLite training and Android side.