Use unbiased stereology method. See publications by Mark J West

There are some points you should consider:

would you like to quantify the number in only the dorsal part of hippocampus or the whole hippocampus. I mean you should consider the bregma level.

Staining of brdu is tricky because you have to make a pretreatment step (HCl treatment) before incubating with primary antibody otherwise you will not be able to detect all the brdu positive cells. this pretreatment affect also the staining of NeuN

good luck

Imam

Several factors may contribute to the different number of BrdU+/NeuN+ cells that you obtained, not least of which is the strain of mouse. Even substrains on the same background may differ. Second is the environment and living conditions. Then of course there are experimental conditions. First, how long did you wait before euthanizing the mice after injecting with BrdU? Typically you won't get very many double labeled cells anyway, and this number is time dependent. You may see fewer cells than the published paper because they collected longer after BrdU administration allowing for more new neurons to mature. Second, as mentioned before BrdU requires antigen retrieval so you may not be getting full penetration of the antibody. HCl is one option, heat-mediated retrieval with a citrate buffer is another. How many BrdU+ cells are you getting in the granule layer? Make sure your protocol is optimized before doing double labeling.  NeuN can ge tricky, too, especially if using a mouse on mouse antibody. Finally, quantification varies by lab/publication. How did the other group report their numbers? Absolute number counted, number per field, per section, per mm2? You may not be as far off as you think. But as mentioned unbiased design based stereology, which incorporates random sampling and volumetric analysis to generate a population estimate is the gold standard. If you are confident that you have double labeled cells and that your coubt reflects the number and distribution you obtained, then don't worry too much about the number not matching the other publication. There are so many factors that modulate neurogenesis and there is a wide range of reported values out there. 

I'm going to echo what Paul said, Chi. You should really be using unbiased stereology to do this. The basic protocol would be to use disectors with inclusion and exclusion boundaries and count each marker separately. This should eliminate the disparities you are seeing. The Stereology Resource Center (https://www.stereologyresourcecenter.com) in Tampa, FL and Microbrightfield have excellent stereology packages to assist with this.