Synthetic evaluation results

[manu15sd]

Hi, I read from other issues that the differences between the paper results and the results of this repository are given by the different datasets (synthetic vs. real) used for training. It would be much appreciated if you could upload the results for the synthetic training just to compare and know if I'm getting the expected results. 

Thank you.

[LiheYoung]

Do you mean the results of solely trained on synthetic images? If so, you can refer to our Table 5.

[manu15sd]

I mean the fine-tuned metric version on synthetic (Hypersim and Virtual Kitty). Table 5 is not fine-tuned, isn't it?