Promote your data rooting and make your ML projects to flourish
One of the most daunting work to do for a Deep Learning engineer is to setup an entire script from scratch for each framework, since each one presents its peculiarities.
Here I'm talking about the most famous DL frameworks: Keras (Tensorflow), Pytorch, FastAI (I love them all!).
I love DL therefore I want the freedom to use (at least) all of those frameworks with no time wasting. I want to focus on thinking about architectures and results, not coding for data structure preparations.
Introduce cytokinin, still not complete, but ready to work on images. Try it!
Check the tutorials folder to quickly get started.
projfolder$ git clone https://github.com/gianfa/cytokinin
projfolder$ cd cytokinin/cytokinin/
projfolder/cytokinin/cytokinin/$ pip install ./
Below an example of data flow is depicted. It all starts with a data loading into the Data object, then you can export your dataset to your preferred Deep Learning framework.