Open windgazer opened 4 years ago
In theory, if we use tensorflow to learn what a playable character looks like from any angle, we would be able to use that model to classify all the objects found by Yolo (including dice). Subsequent models could also be used and create to recognise which direction a model is facing.
Example of dice counter using tensorflow: https://towardsdatascience.com/calculating-d-d-damage-with-tensorflow-88db84604f0a
Use tensorflow to create models to recognise Playable Characters: