This jupyter notebook trains a donkeycar repository of images to identify corresponding steering and throttle values using fastai V2. I have added a transform to augment image inputs by mirroring the image and complementing the steering value. The whole thing runs in under 10 minutes on my laptop.
cmasenas/fastai_navigation_training
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|