Using SSCMA for Google Colab - detecting a person with Grove Vision AI V2

Описание: In this video I will explore the complete process of preparing an existing AI model from Roboflow universe to be accepted in a format suitable for the Seeed Studio Grove Vision AI V2 module. This is AI image recognition for an edge device rather than doing AI image recognition in the cloud.

SSCMA stands for "Seeed SenseCraft Model Assistant" and is a series of tools to get image recognition models into edge AI devices like the Seeed Studio Vision AI V2.

I attempted to re-creating models that Seeed Studio had already completed and had available in the correct file format for Grove Vision AI V2. While I had success with the 'person recognition' model - that is the one I demo in this video, I had problems with the 'Face Detection' Model and it crashed in Google Colab under the SSCMA script.

Most of the video is going thru the steps in Google Colab from uploading the Person Detection model from Roboflow universe to training the model and exporting it in a file suitable for uploading to the target device, Grove Vision AI V2. I do the uploading in SenseCraft on-line to my own Grove Vision AI V2 module.

My next video in this learning process will be to select an image recognition model that I can use in my own project. The model I will select will recognize different coloured Lego bricks on a Lego baseboard. The coloured bricks will represent rocks on mars and the baseboard will be the surface of mars. Eventually I want to use the Grove Vision AI V2 module as an imaging co-processor to the Xiao Esp32-C3. This processor will help navigate a rover on mars with the assistance of image processing using the Grove Vision AI V2 module.

References:
Grove Vision AI V2 at Seeed Studio - Shop ( https://www.seeedstudio.com/Grove-Vis... )
Grove Vision AI V2 at Seeed Studio - wiki (https://wiki.seeedstudio.com/grove_vi... )
Sensecraft AI ( https://seeed-studio.github.io/SenseC... )
Roboflow - universe models ( https://universe.roboflow.com/otc-z2i... )