bohaohuang/mrs

eval_utils.evaluate() applies softmax on post softmax output when saving confidence maps

Closed this issue · 0 comments

scipy.special.softmax(tile_preds, axis=-1)[:, :, 1])