Skip to main content

Electronica 2018 Pidentifier and Automatic Selfie Camera Build

E2018_SelfieCamera_1438180696dfa63d2dc58f5fb072eb4ce030dd01.jpg

Design, mechanical build and components for the AI-powered Pidentifier and computer vision enabled selfie camera.

In addition to the IoT Blockchain Demonstrators that we built for the RS stand at this year's Electronica, we also made two other units demonstrating the use of the Raspberry Pi Camera plus official Camera. The first of these was used in a “photo booth” scenario and equipped with a classic computer vision algorithm running on the Raspberry Pi.

The second demonstrator used the same base hardware plus an Intel Movidius Neural Compute stick (NCS), enabling it to use deep neural networks (Artificial Intelligence) to identify objects placed in front of it. The Pi Selfie Camera and Pidentifier respectively.

Pidentifier

E2018_Pidentifier_a4dcf8a657143d56ac415a56c50672083e6b78cf.jpg
This unit houses a Raspberry Pi and Pi Camera, two Neural Compute Sticks and the official Touch Screen in the upper section. The dimensions are dictated in some part by the focal length of the Pi camera. Objects to be photographed and identified are placed on the base, so the camera needs to be a suitable distance above it.

Diffused LED lights housed underneath the upper section illuminate the base.

A cardboard frame was constructed initially to ascertain suitable dimensions.

pi-dentifier1_18c1dd1f32646f9cbf18aa617740b974a0eb4627.jpg
In the final version the frame was constructed from Bosch Rexroth 20 x 20mm Aluminium extrusion (466-7219) fastened at the corners with the Corner Cube Kit (466-7433) . The front panel consists of a single piece of clear acrylic sheet, bent in two places and supported on the aluminium frame. A single piece of opaque white acrylic was then added to the lower front section to give some added strength and enhanced visual appeal.

The clear top section meant the Pi and the Neural Compute Sticks would be clearly visible and on display. To make the most of this they were fixed to a small panel of acrylic that was suspended inside the top of the Pidentifier, making them the focus of attention.

Bill of Materials

  • Various 3mm extruded acrylic sheet
  •  3m Bosch Rexroth 20 x 20 Aluminium Struts (466-7219)
  •  6 x Bosch Rexroth Corner Cube Kit (466-7433)
  •  12 x Bosch Rexroth 20mm Strut Profile Angle Bracket (466-7354)
  •  2 x Movidius Neural Network Compute Stick (139-3665)
  •  1 x Rasperry Pi 3 Model B+ (137-3331)
  •  1 x Raspberry Pi Camera Module (913-2664)
  •  1 x Raspberry Pi LCD Screen (899-7466)
  •  1.5m PowerLED White LED strip (786-9023)
  •  1 x Panel Mount DC Power Socket (048-7832)

Automatic Selfie Camera

E2018_SelfieCamera1_1438180696dfa63d2dc58f5fb072eb4ce030dd01.jpg
The selfie camera uses Haar Cascades to detect faces and the user has a choice of superimposed hats and spectacles. Upon detecting a face it announces “say cheese”, counts down from 3 whilst displaying red, amber then green lights, before then taking the picture.

This differed from other units we had made for Electronica since it was housed in an off-the-shelf professional wall mounted enclosure (706-7331) . This needed to be drilled to accommodate the various knobs and buttons. A laser cut template was used as a guide to drill the holes.

piselfie_template_b6a977bba95c87b8702c61065adbbd92ca8d8ed6.jpg

A replacement clear acrylic laser cut front panel was designed that would support the Raspberry Pi, Camera, associated electronics and user controls. These were then fitted to the panel and wired. This included an arc of LEDs to light the subject, an LED light RS logo and a Justboom DAC Hat (122-8179) for improved sound quality.

piselfie_back_of_front_dedbedcb310b66e627c394f89cedd4a4462035a1.jpg

The controls on the bottom of the case were then connected and the front panel screwed in place.

piselfie-finished_da8c49654627b74f4456d45306c4141eee31e432.jpg

Bill of Materials

I have a background in the arts, environmental conservation and IT support. In my spare time I do a bit of DJing and I like making things.