wasidennis
commented
4 years ago There could be still modifications in the architecture, even using the same backbone (e.g., VGG-16). We obtain the oracles as reported or use corresponding architectures in each paper.
Why are the oracle resutls of baselines ( based on VGG-16) in Table 2 (Table 5) different? In my understanding, all these results are trained on Cityscapes dataset using the same full-supervised model (VGG-16), which should be the same. Thank you very much!