Class-Weighted Segmentation Loss

Problem

Compute the mean class-weighted cross-entropy over a segmentation probability map.

Each pixel with target class c contributes -weights[c] * log(probs[i][j][c]), with the probability clamped to at least 1e-12. Return the mean over all pixels rounded to 4 decimal places.

Test cases

case 1

in: [[[[0.8,0.2],[0.1,0.9]]],[[0,1]],[1,2]]

expected: 0.2169

case 2

in: [[[[1,0]]],[[0]],[3,1]]

expected: 0

case 3

in: [[[[0.5,0.5],[0.5,0.5]]],[[0,0]],[2,0]]

expected: 1.3863

Starter codePython

def weighted_seg_loss(probs, target, weights):
    # Your code here
    pass

3 test cases · run your code against them in the editor