I've been working with CNNs recently. For a new task, I need to predict objects in an image pixelwise. Fully ConvNets seem to be the way to go. I read the original paper (Long et al., 2014) and a few blogs about FCNs so far. Though, I have some trouble understanding how to setup my training data.
- From what I've read, each image in the sample data can be (in opposition to CNNs) of arbitrary size, is that correct?
- To setup the training data, I have to create a second "image" of an image, which should act as a mask for the object to be trained. May this mask-image contain more than one object and also different types (classes) of objects?
Thanks in advance. Please let me know, if I need to add more information.