Hello,
I downloaded the Cityscapes dataset and provided pretrained Models according to the readme. When I run evaluate.py the mIoU scores are different to provided ones:
train: 67.26%
trainval: 80.90%
train_bn: 67.30%
trainval_bn: 80.80%
Hey,
According to my experience, the mIoU scores may be a little different due to different tensorflow version, but it will vary in a reasonable range.
Hello, I downloaded the Cityscapes dataset and provided pretrained Models according to the readme. When I run evaluate.py the mIoU scores are different to provided ones: train: 67.26% trainval: 80.90% train_bn: 67.30% trainval_bn: 80.80%
Does anyone else get the same results?