In January of last 2021, my team created a series of projects documenting how to train a model to play the card game Dobble.
It walks through all of the steps of collecting data, training the model in keras, testing the model, augmenting the dataset, and finally, running the model on Avnet's Edge AI hardware.
I've compiled links to our projects so far:
|
This project explains how the math in the Dobble card game works. It will also give you access to the dataset, walk you through installing the required libraries, and run a script to look at the dataset. |
|
|
In this part, you can train your own Dobble-solving model using Keras. To improve the model's accuracy, we use keras' built-in data generator to augment the dataset. |
|
|
Run the Dobble Challenge on MaaXBoard In this project, you'll finally run the card game model you trained on MaaXBoard! |
|
|
Deploying the Dobble Challenge on the Ultra96-V2 In this project, you'll convert the model you trained to .xmodel format to run on Ultra96! |
|
