Add upsampling layers to existing model

[GGDRriedel]

trafficstars

Regarding this:

https://github.com/divamgupta/image-segmentation-keras/issues/110

I would like to add upsampling layers that would make me able to match my data.

I haven't found a way to.

My first try was to import the layers.Add method from keras, hoewever, since the model is already initiated with weights, this is not possible. Neither is doing this before initiation since I would get a weights missmatch.

I tried

model4=resnet50_pspnet(n_classes=2,input_height=960, input_width=1152)
from keras_segmentation.models.model_utils import get_segmentation_model, resize_image
from keras_segmentation.models.config import IMAGE_ORDERING
model5=resize_image(model4, (960,1152), data_format=IMAGE_ORDERING)

But that doesn't work either. I would be really thankful for a poke in the right direction.