Scoreboard data extraction pipeline using OCR

Valorant Scoreboard OCR and Data Extraction
This repository contains scripts for extracting, processing, and parsing the scoreboard from Valorant game screenshots. The project leverages computer vision techniques with OpenCV and optical character recognition (OCR) using Tesseract to extract in-game statistics such as player names, KDA, and CREDS.

Project Overview
The project is designed to work with Valorant scoreboard screenshots. It consists of three main stages:

Cropping the Scoreboard Sections:

Red Section:
The red.py script crops a specific rectangular region from the lower part of the screenshot, typically capturing the red scoreboard area.
Green Section:
The green.py script crops a rectangular region from the upper part of the screenshot where the green scoreboard appears.
Masking Unwanted Data:

Green Section Masking:
The greenmask.py script processes the cropped green scoreboard image by blacking out specific columns that contain extraneous data. This step refines the image for improved OCR performance.
OCR and Data Parsing:

Preprocessing & Extraction:
The green_final.py script preprocesses the masked image (converting it to grayscale, resizing, and thresholding) to enhance OCR accuracy. It then uses Tesseract to extract text from the processed image.
Parsing OCR Results:
Extracted text is parsed to obtain structured information (player names, KDA, and CREDS) and output in JSON format.
Project Structure
7.png
A sample image that is a Valorant scoreboard screenshot.

red.py
Crops the red section from the lower portion of the screenshot and saves it as red_cropped.png.

green.py
Crops the green section from the upper portion of the screenshot and saves it as green_cropped.png.

greenmask.py
Masks unwanted columns in green_cropped.png to produce a cleaner image (green_masked.png) for OCR processing.

green_final.py
Preprocesses the masked image, extracts text using Tesseract OCR, and parses the text into a structured JSON format representing the scoreboard data.

Dependencies
Ensure you have the following installed:

Python 3.x
OpenCV (opencv-python)
Tesseract OCR and the pytesseract package
Pillow (PIL)
Install the Python packages using pip:

pip install opencv-python pytesseract pillow