IBM announced in early 2016 that the artificial intelligence (AI) of its Watson AI supercomputer had acquired image recognition capabilities. Starting today, you can try it out for yourself!
A site connected to IBM Watson lets you feed the computer your own photos to see what it thinks is in them. The results are impressive.
The visual recognition demo allows you to give Watson an image URL or upload a photo. The results come back a little later (a few seconds) and display what it thinks it “sees.”
On the test page, there are already uploaded photos, but you can also upload your own, as mentioned above. Watson's test with some random photos that were already on the site was quite accurate. It could understand what was a landscape, an animal, which animal, and even what was in the background.
From the same page, you can also train Watson by uploading photos and making a rough classification. This way, it can learn what it “sees” in that particular classification. You will need to upload more than 50 photos for it to be able to reliably say what it “sees.”
It should be mentioned here that the supercomputer's lifelong learning will not be based only on your own photos. The system will cross-reference data with every new photo it doesn't already have in its database. This way, it will become smarter over time.
The demonstration technique is available as an open source project. So anyone interested can clone it and run it on their own system….
See the photo recognition page
https://visual-recognition-demo.mybluemix.net/
