AI on the edge

This is a housing for an integrated AI system for processing image on a single device, that is very cheap (less 10 EUR) and very small. The overall system is running a software environment, that combines image aquisition (incl. light source), image preprocessing, image recogniztion with neural network systems (keras / tensorflow / tflite) and a web server with API for setting and data readout.

The software including electrical setup and configuration is described here: https://github.com/jomjol/AI-on-the-edge-device

Additional to this housing the following components are needed:

• ESP32-CAM
• micro usb connector (optional - 5V power cable also sufficient)
• SD-card for software

This Thingiverse description is only about making the housing. The major task was to make the housing as small as possible including the housing and access to the SD-card slot. Therefore beside the cover and the top, an inner part is designed, that support an integrated micro USB port for 5V power supply. To use this device in applications a clippin solution is suggested, therefore cavities on the two short sides are integrated.

One example of usage will be an updated version of a watermeter, the hardware update for this will be published soon.

The inner part for the micro USB connector might need to be adapted to the individual solution. In principle a 5V and GND cable is sufficient.
The cover has a slit for accessing one side of the 8 pins for accessing the flashing pins (details about the wiring see also link above).
Additonall a hole is going through to access the reset pin on the board. The pin to acces this is made out of a about 1 cm part of filament, which was flatened at one end to prevent sliding out.