Misclassification models are available in unmarked R package or Presence. Any of these programs/methods should provide site-specific occupancy estimates with Presence being the most user friendly. Bayesian versions are also available (see Chambert et al. 2014 in Ecology; supplementals contain WinBUGS scripts).  See links for more info. 

https://github.com/rbchan/unmarked/blob/master/man/occuFP.Rd

http://www.mbr-pwrc.usgs.gov/software/doc/presence/presence_doc.html#falsepos

http://onlinelibrary.wiley.com/doi/10.1890/14-1507.1/full

Hi. In addition to Brian post, as you know, false positive error rates can induce extreme bias in estimates. I recommend you these paper:

https://www.uvm.edu/rsenr/vtcfwru/spreadsheets/occupancy/Occupancy%20Exercises/Exercise12/Royle_Link_2006.pdf

http://onlinelibrary.wiley.com/doi/10.1111/1365-2664.12303/abstract

Thanks Brian and Jose for your response. Though any of the Miller models (2011) or recent Bayesian models can estimate false positive probability and site specific occupancy, my problem is that I do not have site-specific covariates.The sites I am dealing with (villages only) are fairly homogenous in habitat. I was hoping that there might be a way of using the overall estimated false positive detection prob. (which I have obtained from analyses) to correct the raw detection prob. at each site.