Use pictures taken by the Vision AI Dev Kit to train your model

What you will do

  • Use Vision AI Dev Kit to take pictures you will later use to train your Vision AI model

What you will need

Take pictures with the camera

  • Open command prompt that allows you to run python scripts. When using Windows command prompt add python to the path using a following command

    set path=%path%;[path to your python folder]

  • Go to the directory in your device where you have the contents of vision-ai-developer-kit GitHub.
  • Go to folder vision-ai-developer-kit/samples/research/ai-vision-devkit-get-started-legacy/modules/VisionSampleModule/python_iotcc_sdk/sdk/

Run command:

 python .\ --ip {IP address of your device} -t {object of interest} -p False -n {number of pictures}

For example:

 python .\ --ip -t Apple -p False -n 10


  • IP Address is the IP address of your device
  • Object of interest is the object you are taking pictures from like “Apple”. The script will create a folder with the same name to the directory where the script is located and store the picture to the folder
  • Number of pictures defines how many pictures the camera takes. The time between pictures is roughly 1 seconds

Note: it may be that taking pictures with the camera causes the VisionAIDevKitGetStartedModule not to function properly afterwards and the device needs to be rebooted in order to get the module to inference normally.