Many comments here are sharing their results from using the examples. It's important to keep in mind the examples are trained on datasets that may or may not include data on the examples you are uploading, which means random images or words not included in the training data sets will likely return bad results.
Please keep this in mind when judging the quality of the library....the data you use to train the model affects the quality of your results.
It's night over Toronto from the station with lights glowing. I imagine it got it from that or something. I think it was a Canadian Space Agency photo, but I can't remember the exact source/which astronaut.
I thought this would be blowing up HN, but eh, tumbleweed here!
Is there some reason the HN community isn't too gung-ho about this? Is it because it's in the browser or is it javascript or that it's too high level (vs low level details). I'm completely Stumped by this response.
Portrait pictures of myself:
* academic gown, academic robe, judge's robe 38.01% confidence.
* Band Aid 88.27% confidence.
* Prison 42.74% confidence.
* Wool, 11% confidence
Moving on to Random stuff:
- Linkedin Logo -> Hair Spray
- Washington Monument -> Pole
- Abraham Lincoln Statue -> Altar
- Dragon Fly -> Kite
- Sun Glasses -> Car Mirror
- Mushroom -> Mushroom!!!!!!
- Lipstick -> Lipstick !!!