I am a new deep learning practitioner and I started my journey with fastai. I came across a problem where I was supposed to pass pixel data with RBGB channels to databunch in order to create a cnn model. I googled a lot, but could not succeed. Does anyone of you know a workaround this problem? Thanks!
Do I understand you correctly? You have a .csv containing RGB pixel data?
That’s a horrible way to store images, you should write a script to convert this .csv to actual image files…
Yes, you got me right. I understand, but that is how the data has been provided for a challenge that I am working on. By any chance, you know how I can convert this data to images?
Well that completely depends on how the image is stored in csv. If you provide a sample csv and the according information from the challenge, I can take a look.