Sorry. I don't understand your table. What are the Mean values?
Baseline Count per Unit is your estimate of what is the typical counts per unit. This is required as part of the power analysis. When you have the baseline it is then possible to determine the power to detect a given difference from that baseline for given sample size. Or you can calculate the required sample size for a given different to detect and power (as you said).
I suggest you try different estimates for Baseline and look at the effect on the sample size estimate.