Does having a tf.nn.dropout layer in a tensorflow frozen graph affect dnn::forward

edit

I wanted to know if frozen tensorflow graph ( loaded using dnn::readNetFromTensorflow ) containing a dropout layer affects dnn::forward i.e. are the neurons from the previous layers dropped at random or is the dropout layer omitted during the forward pass.