model naming scheme
{project}-{dataset}-{encoder}
{openfibsem}-{baseline-dataset}-nc{num-classes}-{resnet34} -> openfibsem-baseline-34.pt {autolamella}-{dataset-02}-{resnet34} -> autolamella-02-34.pt
all models are num_classes = 3 unless specified
when calling load_model you should use these as so:
from fibsem.segmentation.model import load_model
# load openfibsem baseline model, with resnet 34 encoder, num_classes = 3
model = load_model(checkpoint="openfibsem-baseline-34.pt", encoder="resnet34", nc=3)
# load the latest autoliftout model, with resnet 34 encoder, num_classes = 5
model = load_model(checkpoint="autoliftout-baseline-nc5-34.pt", encoder="resnet34", nc=5)
