5π Consulting

Uploaded thousands of Photos to Google Photos, here are the best classification mistakes

September 02, 2015

I’m a lazy person who never took time to organize their photos so I got really excited about Google Photos and just finished uploading all my Photos to it. I used flickr in the past but just dumping photos there, even if I can search for EXIF tag data doesn’t make much difference to storing them on a hard disk in a drawer.

Google Photo on the other hand understands what’s going on in a picture. Google uses a neural network which is trained by millions of photos on the web to identify objects in pictures.

This works very well and when thinking about it’s mistakes, the uncanny feeling just amplifies: They are mistakes, they aren’t bugs. It’s not faulty, it’s just a little unexperienced - almost like a child. In fact, I would be very interested in a catogorization standoff between a child and Google.

#Concerts Screen Shot 2015 09 02 at 14 42 18 I’m really not sure about this one.

#Dessert Screen Shot 2015 09 02 at 14 44 38 Let’s finish this course with a nice bowl of raw chicken.

#Dogs Screen Shot 2015 09 02 at 14 40 14 To be fair, you could consider british shorthair the dogs of cats.

#Birds Screen Shot 2015 09 02 at 14 39 47 Birds. That’s definitely correct.

#Skylines Screen Shot 2015 09 02 at 14 49 11 Bar graph or 8-bit skylines?


Johannes Ziemke

By Johannes Ziemke.
Cool, cool but need help? You can hire me!.



Impressum