Google and NXP advance artificial intelligence with the Edge TPU

46 thoughts on “Google and NXP advance artificial intelligence with the Edge TPU”

1. 1:30 Google talking about Privacy!!

2. nice, javascript all the way

3. Pricing? Debian support? Is there a 2+GB RAM version? eMMC? I need more specs!

4. Amazing

5. ice cream or hot dog !!!???
   Will this board be affordable?

6. price please. when is going to be released to the market

7. What differentiates this from let's say a Raspberry pi and opencv?

8. This look like powerful respberry. Pi. Hope google will make it affordable and available to countries like Nepal too.

9. Fpga?

10. so basically a raspberry pi with a tensorflow optimized mini GPU.

11. ??

12. What is Google?
    Can anybody explain me

13. Google caring about privacy?
    Cmooon dudes

14. They are showcasing AI versions of Arduino projects.

15. The "bounding boxes around faces" is a few decades too old to wow anyone. But server-less (good) voice recognition & natural language processing could be a game-changer.

16. What is this?

17. Now you can cook something and ask it what it was …

18. Did he say WiFi?
    Should we be concerned? LOL

19. I was thinking adding TPU usb sticks to raspi, then google made its own raspi

20. So forget the raspi if you are doing a tensorflow ML project and switch to this new device as soon as you can

21. We teach our kids rules and what's right andwrong. How do you teach AI that, and furthermore how do you discipline it. Are we playing with fire without understanding its capacity?

22. It's now actually EASY to program that on a raspberry pi.
    Maybe it would be a bit slower but results would nearly be the same.

23. There was a glitch.

24. And here comes the matrix

25. so its a Raspberry Pi board with a Tensor flow gpu ?

26. so… hot-dog, not hot-dog. got it

27. when the orange donuts was shown to the system, before training on it, it was wrongly classified as an ice cream (red light)….False positive 🙂

28. How about a free giveaway?
    I'm waiting.?

29. google: "ML on the edge can be used for smart speakers, smart dishwashers and smart TVs"…

30. Dude please learn how to make a video

31. How far off is this ?

32. Very Cool?️‍?️?

33. in 10 years devices from this video will be famous memes

34. Need a deep autoencoder for MNIST

35. More junk for sale.

36. not the ai I was hoping for

37. Wast of time

38. I really like the Demo at 2:24. What is your approach to differentiate between the objects? Is it online transfer learning or a siamese approach?

39. I was ready months ago. What's taking so long to release??

40. privacy
    fool me twice

41. CHEEZZZZZEBURGERRRRR!!!!

42. Message in a bottle: I wish you told us some specifics about how large a model we can load the TPU with, that kind of things. Basically could not find any kind of information regarding the TPU's capabilities. I understand the concern for IP and such, but please at least give us functional specs.

43. You just killed a few start ups 😛

44. I already found a cool application for this. Thank you for sharing!!!

45. very cool!

46. I was wondering if they had any audio recognition examples

Comments are closed.