Need a clarification about the how much of data were supplied to retinanet model_network by giving batch_size, epochs and steps per epochs.

So far, I thought that the steps size was calcuated as following:

step_size = (total_number_of_data/batch_size)*epochs

Whereas in keras-retinanet, it takes batch_size, epochs and steps_per_epoch as arguments, which was different from above mentioned case. I have a doubt how the calculations takes place? Thanks in Advance