intel/handwritten-chinese-ocr-samples

troublesome Installation

kucingkembar opened this issue · 2 comments

kucingkembar commented

hi, sorry for my bad English, and sorry if I rude,
I average windows user and tried to install this, and there are too many steps and I don't know what I must do with the programming code,
so, can I ask you simpler way or "portable" version of your program?

sorry again for my rudeness, and thanks for reading! have a nice day

bliu3650 commented

@kucingkembar Thanks for your question. The best option might be providing docker image for the users who came here and wanted to have a quick trial, but I've switched to another organization so I could not update this repo any more.

For instructions to reproduce this work, some suggestion are:

  • if you would like to deploy the pre-trained model, you'd better to start from the Intel OpenVINO toolkit;
  • if you are going to train model with your own data, you need to get familiar with the training procedure with the PyTorch imagenet example.
kucingkembar commented

first, thank you for the reply,
second, I tried another OCR like FineReader(paid version, blank output), tesseract, and easyOCR (the accurate about 95% reading Chinese computer font), your program seems promised
3. the installation alone is quite nightmarish, and I don't want to train, I just wanna use it
4. other python "program" when I install using PIP, they usually have EXE, are this "program" do not have it?

thanks again for reply and have a nice day