Closed AshinCarlXiao closed 5 years ago
We set no limitations on the max flow that the model can predict. It probably depends on the distribution of the training dataset. You can check out the supplementary material of this paper https://arxiv.org/pdf/1504.06852.pdf
Is there a limitation of max flow that the model can predict? Specifically, if the motion is lagger than a value(maybe the largest flow of the training set), the model still can give a good prediction? If there is a max flow that the model can predict, can you tell me the value of max flow?