Python Tennis Graphic Extractor

I have a folder full of broadcast-style tennis stat screenshots. I want a Python script that will read every image in that folder, recognise the on-screen text with very high accuracy, parse it, and drop the results into a single Excel file saved to my chosen output directory.

I’d like every single field that appears on the graphic—if the image shows aces, double faults, first-serve percentage, names, country, tournament name, time, or any other stat, etc, that should be included as a column as well.

Key points for the build
• Folder-based batch processing: point the script at an input directory and it should auto-process every screenshot it finds.
• High-accuracy OCR: Tesseract, EasyOCR, PaddleOCR, or another library is fine as long as the results are clean. Pre-processing with OpenCV (de-noise, threshold, perspective correction) will likely be needed.
• Robust parsing: the graphics aren’t always positioned in exactly the same place, so coordinate-based cropping or template matching may be necessary.
• Output: one neatly structured .xlsx file with each match on its own row and each captured field in its own column.
• Self-contained: everything should run from a single command, with a requirements.txt and clear usage notes so I can recreate the environment quickly.

Acceptance test
I’ll provide a small sample set of screenshots. If the Excel file produced matches what is visible on those images with minimal manual correction, the job is complete.