Hackaday Prize Entry: Pi-Driven Google Glass

[Ricardo Ferro] didn’t want to buy a Google Glass, so he made his own.

The Raspberry Pi Zero Prism consists of a 3D-printed headset the side-pieces of which hold a variety of electronic components, including a Pi Zero running Raspbian Jessie, a Pi Noir IR camera, a WiFi/Bluetooth module and a whole mess of SMD tactile push buttons. Video output is provided by a Kopin 922K display module. This module is usually used in smart goggles and uses a prism to reflect information into the wearer’s field of view.

One application [Ricardo] envisions for this Open Source Google Glass is using it in conjunction with facial recognition software and the YouTube-favorite IR camera trick of seeing through clothing. No, he’s not using it for that idea, and you should get your mind out of the gutter. [Ricardo] wants to identify masked criminals. Setting aside the technological challenges of making that technology work, we think that walking around with x-ray specs is likely to get those specs broken off your face by someone who wears clothes for modesty purposes. Still, it’s a fascinating project and we love the way the prism and video assembly comes together.

11 thoughts on “Hackaday Prize Entry: Pi-Driven Google Glass

  1. It looks like a lefty. I always wear my rear-vue mirror on my glasses on the left because of right side driving in USA. I still get the occasional comment “is that Google Glass?’ no because GG is on the right side. We are mostly right handed and right eyed for info processing.

  2. “someone who wears clothes for modesty purposes”

    Probably a small group…. Seriously though, does anyone have source on these as individual modules? A quick google led me to finished products or modules that were no longer available.

  3. I am genuinely curious here, the MotionPie software runs pretty awful on my Pi zero and Pi 2 even with ethernet vs wifi. The FPS is like 4 -5 fps, its basically unusable for video.

    What exactly are you running regarding video recording software, etc?

    1. Not the original poster, but I had great success with this software to run my Pi camera:
      http://elinux.org/RPi-Cam-Web-Interface

      Even on a Zero, it was quite responsive. The built-in motion detection routines (NOT the stand-alone “Motion” software) run in the GPU and run essentially at full frame rate. Very low latency, too, all things considered.

      You need a fast SD card to write the mjpeg video stream to disk. With a Class 10 card, I had no problems writing the stream to disk, as long as nothing else IO-intensive was happening. (The software can be configured to use MP4Box to re-encode the video to mp4 for streaming — if it’s trying to re-encode video A while simultaneously recording a new video B, expect dropped frames in your new recording.)

      I used it to build a little remote IP cam with onboard motion detection and local storage of motion-detect events. Ended up not really using it, though, because my primary interest was for keeping an eye on things at night, and it turns out that the pi cameras are completely AWFUL in low-light conditions. (Yes, I tried the NoIR version and it still sucked even with IR illumination.)

  4. I like how you say “identify masked criminals” and not “narc on protestors”.

    Today’s savvy cyber punk is going to have to wear cv dazzle to protect her anonymity, and thick clothes to protect her modesty :P

Leave a Reply to PaulCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.