One of the biggest advances in this early stage of working artificial intelligence is facial recognition — the ability of Facebook to pick out individuals across its platform, and the Chinese to nab suspected criminals out of a city crowd.
But last year, engineers working for JD Finance, the financial arm of JD.com, the e-commerce giant, wondered where such capabilities could go next. What about animals? So it was that a contest unfolded: three teams of engineers pitted against one another to adapt facial recognition programs to piglets.
How will the winner be selected?
- As a first stage, each team must show it can identify individual pigs, says Chris Yao, a vice president of strategy for the JD.com affiliate, speaking over lunch Monday with a small group of journalists.
- Then, they must do so over time — after all, as with humans, a piglet's facial appearance evolves as it grows up.
This isn't entirely a lark: The pig inquiry has its roots in a prior JD Finance study of chickens that resulted in a way to systemize feed costs for a full-size bird.
- If facial recognition can detect a pig's age and approximate weight through time, it might be possible to systematize the cost of feed to raise them, too, and "when it's the right time to sell," Yao said.
- Yao said he has begun to look around for who might pay for such data.