image-segmentation-keras
image-segmentation-keras copied to clipboard
`class_weight` not supported for 3+ dimensional targets. Issue with python3
trafficstars
I am using keras segmentation to detect paths and the rest of the terrain, the pixel percentage of the route is much lower than the terrain pixels. So I calculate the percentage with this function.
The problem is that with python2 works but with python3 doesn't.
The function returns a dictionary:
weights_dict ={0: 0.11, 1: 0.88}
I don't know how to solve this.
