In our DAS 2016 work - 'Multilingual OCR for Indic Scripts', a comparison of the popular Devanagari OCRs with our OCR is presented. A dataset comprising of 100 scanned images of printed documents were used
the dataset used is available here
If you are using this dataset, please cite our paper
  author    = "Mathew, Minesh and Singh, Ajeet  and Jawahar, C.~V.",
  title     = "Multilingual OCR for Indic Scripts",
  booktitle = "DAS",
  year      = "2016"