I'm preparing a dataset for a multiclass semantic segmentation using U-Net like architecture. To be precise, I've got it ready but a question came to my mind. How does pixel values of a segmentation map influence the training? Also, is it better to have a greyscale segmap, or RGB one?

I have the dataset labeled and augmented, the only thing I am thinking now is if I should alter the segmaps.

I am planning to use Keras, is it smart enough to take in segmaps in both forms? I haven't found an answer anywhere, doublechecked if not on this forum, I hope it's not something super trivial. First time trying to do segmentation task so it's all a bit new to me :)

Edit: right now the segmaps looks like:

  • background [0 0 0]

  • object_type1 [16 180 75] green

  • object_type2 [225 225 25] yellow

  • object_type3 [230 25 75] red


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.