Here is my first attempt to get Vision_UI (Visual GUI) to work on Google Colab (not with the same functionality…yet). Only works with data in folders but the aim of Vision_UI is to provide a usable graphical interface.
This version for colab mounts the drive and provides an easy to use ‘copy and paste’ method to set the path and choose the image for augmentation.
You can review a number of parameters (GPU, TPU, number of files, choose your NN, pretrained settings, normalization, batch size, image size, various augmentation parameters, view batch, select metrics, view LR and train your model which is saved in the models folder of your path.
All within one line of sight and running two code cells.
There are currently some visual challenges that I am working on when comparing this to the version that can be run on my computer and it is a work in progress. Also the results tab has not been included as I am working on getting it to work better before posting it. It would be great to get some feedback on this.
