Rubber Duck or Cow?
Numenta recently released new demos for download on Mac or PC. Vision4 demonstrates how the Hierarchical Temporal Memory (HTM) technology can be used to recognize photographs of given objects.
After training the HTM network by showing it 3500 photos of 4 different types of objects: rubber duck, cow, sail boat, or a cell phone, the HTM network is able to accurately predict what it’s looking at. You can even show it a picture of a cow that looks like a rubber duck and it will take an educated guess. Er, have a look to see what I mean:
The demo will not correctly recognize all images. Sometimes this is because the network has never seen a similar image. For example, the network was trained with pictures from the front and side of a cow, but not from the rear. If you show it a picture of the rear of cow, it may not recognize it as a cow. In other cases, the reason that the network makes a mistake will be less clear.
For sites still using visual captchas to keep out fake robot visitors, time to re-think your strategy! We also can’t wait for TinEye to get their hands on this technology. Who knows what incredible applications they could build with it!
Recent Comments