New Bird App

So I’ve been continuing my machine learning adventures, and I’ve recently published my first Google Play app! (Sorry iPhone users)

The app can identify up to 200 bird species (I used the Caltech Bird Dataset), and image classification is done with a convolutional neural network (a bunch of nerdy stuff). For anyone interested in how it was made: I used Tensorflow (A really nice machine learning platform made by Google) to train the model and deploy the app.

My review of the app

The app is pretty good at identifying the TYPE of bird (like a sparrow, raven, etc), but it isn’t that great at identifying the exact species. However, there some nice species that the app can identify like a red-winged blackbird or the cape starling. I have so far only tested this on Google images… birds are pretty hard to find around here and even harder to capture on camera so… that is one downside that I regretfully admit. But hey, the app works! And I’m happy about that ^.^

Further notes

I have more machine learning projects planned! This is all new to me, so I’ve been getting stuck here and there. Hopefully, I can make some progress this week.