Preparing ML algorithm for embedded application:

Choose a dataset from Kaggle or any one of the datasets that I have already provided

Train them on MobileNet, the foundation computer vision model that is available as opensource. You many use Edgelmpulse for this purpose

Test the model and provide the confusion matrix

Use Pruning and Quantization techniques to optimize the model for MSP $32$ or any other embedded device you choose including smart phones

Recommeded to use

edgeimpulse.com $($ you can also use any other equivalent tool$).$ Submit the model you developed, the name of the embedded device you optimized the model for and the dataset you ujsjs