OCR for TotK coordinates
- ocr_rt.py - Captures, OCR, and plotting in a single script (assumes a mac)
- The process is slow due to OCR, it may a few frames / second
- ocr_file.py - Dumps xyz images for use in OCR from a video file
- Usually run these through Tesseract afterwards using gnu parallel, then post process
The pre-processing of the images assumes the minimap is focused on the Sky for a consistent background. Be aware any movement into the menu will likely change the minimap to the current level.
Using a different level minimap (Depths or Surface) would likely require different cv.inRange()
parameters or other pre-processing.
Doing this is realtime ocr_rt.py is slow. I have found that circumn-aviagating the rooms by climbing
on the walls does a reasonable job of capturing the outline with the least amount of data. Running around
the room might be ok but you would likely want to use the ocr_file.py for a higher details.
- OpenCV for image capture and processing
- Tesseract for OCR
- python3
- PIL in ocr_file.py for image conversion
- pytesseract in ocr_rt.py for OCR
- matplotlib in ocr_rt.py for plotting
- numpy
BSD 2-Clause