Michelle Chismon - Anything is possible!* Learning with AWS DeepLens *But not necessarily easy...
- 1. Anything is possible!* Learning with AWS DeepLens Lenny Chief DeepLens Michelle Chismon Technical Solutions Consultant *But not necessarily easy...
- 3. Why are we all here? Learn something new! Learn about Machine Learning... ...And how it relates to the AWS DeepLens
- 4. The Journey
- 7. But where to begin?
- 8. AWS DeepLens
- 9. Camera Camera Status Wifi Power Status Power Button 47 mm 94 mm MicroSD MicroHDMI USB x2 Audio Out Power *Dashing debonair look not included... *
- 10. AWS Integration
- 13. Machine Learning “Machine learning algorithms use computational methods to “learn” information directly from data without relying on a predetermined equation as a model.”
- 14. Examples
- 15. Supervised Learning Dog Dog Dog (or polar bear)
- 18. Reinforcement Learning ● If car is within 10% of the centre line, give 10 points ● If car is between 10-30% of centre line, give 5 points ● If car deviates further than 30% of centre line, give 0 points
- 19. Pictionary Time!
- 20. Quick,Draw!
- 21. Tensorflow Open Source Allows layered architectures Can be run on pretty much any system Visualisation tools Widely supported with a good community
- 22. Failed, back to the drawing board
- 24. Cats, Rats, AI, Oh My...
- 25. Bird Classification >11,000 images, >200 species
- 26. Bird Classification Reformat images Define hyperparameters Define model Train model
- 28. The Journey 30 29 28 27 26 25 19 20 21 22 23 24 18 17 16 15 14 13 7 8 9 10 11 12 6 5 4 3 2 1
- 30. Reality
- 31. Troubleshooting When you get an error in your DeepLens execution and Google has no answer
- 32. Device Offline
- 33. Invalid Password
- 35. Over to you Lenny!
- 36. Things I’ve Learned ● It isn’t easy to adapt Tensorflow when you have no knowledge of Tensorflow ● An inanimate object can test my resolve a lot more than I had anticipated! ● The DeepLens has huge potential and with its tie-in to AWS services, could be a powerful device!
- 37. Thank you for listening! [email protected] beaumont-michelle https://nubego.io nubego [email protected]