i was trying to make a reinforcement learning agent that can learn to aim and shoot in an online shooting arena, i wanted to extract the score card from the game window for the reward function. i have been using tesseract ocr but it is continously failing to detect digits accurately it detects 1 as l ,7 as / ,3 as B and sometimes some other symbol for the same number. i have spent much times in preprocessing the frames but the ocr model seems to perform worse.
Related Topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Seeking advice on selecting the best OCR model for business card recognition | 1 | 510 | April 23, 2024 | |
AI for survey scoring | 1 | 311 | January 8, 2023 | |
Training a model for a PDF with OCR - where to begin? | 3 | 7454 | December 13, 2023 | |
Looking for OCR post-processing for Visual Document Understanding | 0 | 471 | December 15, 2023 | |
Image to text models tailored for web scraping? | 1 | 196 | June 9, 2024 |