What all inputs needed to finetune on a new dataset?

hellochick / PSPNet-tensorflow

TensorFlow-based implementation of "Pyramid Scene Parsing Network".

326 stars 123 forks source link

What all inputs needed to finetune on a new dataset? #33

Open 314rated opened 6 years ago

314rated commented 6 years ago

Hi, I have a dataset of ~ 1000 images and corresponding segmented images. Number of classes in these images = 1 (ie 1 object class, and rest is background). Are any inputs in the form of coordinates also needed, or can one finetune using just image-segmented image pairs? Could you also present the steps to finetune on a new dataset?

Thanks