Skip to content

alvarowolfx/gcloud-iot-edge-tensorflow

Repository files navigation

Asset Tracking using Edge Computing and Computer Vision.

A bunch of ESP32 with camera, a local server (probablye a raspberry pi ) running image classification and object detection using Tensorflow and sending the data to processed data to Google Cloud.

[Work in Progress]

Upload firmware with PlatformIO

Open esp32-camera-firmware folder on PlatformIO. Now the firmware have support for two models of esp32 with camera:

  • ESP32 Cam from M5 Stack
  • ESP32 Cam from SeedStudio

Depending on your model, change on the platformio.ini file the env_default configuration depending on your board (m5cam or esp32cam). Also you need to change the Wifi credentials on the sdkconfig.h file ( CONFIG_WIFI_SSID and CONFIG_WIFI_PASSWORD).

Then click on upload to flash the firmware into the board.

Run server edge node

The server was written using NodeJS, Tensorflow.js library and the CocoSSD model to detect objects on the image.

Run the following commands inside the edge-server folder to setup the server:

  • Install dependencies:
    • npm install
  • Run server:
    • npm start
  • Open localhost:3000 to see the UI

Google Cloud Setup

[Work in Progress]