Amazon DeepLens first impressions

Submitted by isntall on Thu, 11/30/2017 - 19:10

(I haven't actually deployed anything to this device)

I went the training for this device, it was wildly popular. I was fairly impressed how easy the device was in the lab. It was able to recognize humans and a hotdog (worked better with ketchup); all part of the labs. Great.


I wanted to get something working on this device when I got back to my hotel. So after some unboxing shots, I followed the getting started material.


I plugged it in got it connected to the devices wifi and went thru some initial setup.

And then wait. The info said it should take 5 minutes. 10 minutes go by, 30 and then 60 at which point I pressed the power button; this could have been due to poor internet. The device did not come with a micro-HDMI cable so I couldn’t see what was happening. Round 2 and it work much faster.


I installed the cert and off to the races.

The docs recommend the user disable ssh and use the AWS Console, but with no video out I have almost no connection to this device; besides ssh is how I do everything.


So no I have this device the forum is almost non-existent (atm) and info is scarce outside of the prescribed lanes at least on the AWS docs side.


The package comes with a DeepLens, power adapter, 32G micro SD card; mine had a workshop credit too, probably not standard.


I left auto updates on, but they have not ran the Ubuntu 16.04 OS upgrades, just the greenfield updates; via the update.service


It does not look like the device auto provision the micro SD card.


This is still a very new product and it has a lot of potential; especially nice because you can the video on the device and only pass the results/metadata or run it completely offline.


And for the moment mplayer does not play nice with X forwarding to my mac.


A couple more takeaways:

  • user: aws_cam
  • power supply 5V 4A


If you get one of these order a micro-HDMI cable now.