Well, if you were building an app to recognize sign language, that would take a lot of recognition work, particularly if you weren't training for one of the major sign languages, such as ASL. There is a large and ongoing debate in the ML world about whether it's harder to improve algorithms or get more training data, and in my view the latter is harder by far - anyone can write code and throw 10,000 GPUs at it, but getting 100 different photos of the same hand gesture, with varying skin tones, lighting, angles, and more, is hard. Fortunately for all of us, sites like Kaggle share lots of existing training data, which is a good start.

Ah, I see it in the video now, (and was pointed this way by the dev forums) but it needs to be a macOS playground, not iOS 😁

sorry. It was a sarcasm. I wrote my comment too early. 5 of 5