/ml_game_score_extraction

App for extracting game score and participants in FIFA19

Primary LanguagePythonMIT LicenseMIT

Game score extraction

License Python 3.7 Tensorflow 1.15 Solvve

Description

This application allows extract FIFA game score from photo or screenshot.

Score extraction is based on searching region of interest and text recognition from the region:

{
  "score":  "JUVENTUS 3 - 2 FC BARCELONA",
  "team1":  "JUVENTUS",
  "score1": "3",
  "score2": "2",
  "team2":  "FC BARCELONA"
}

Installation

To use the application you need pretrained weights and language data for Tesseract-OCR - please, contact ML team for more details.

  • Clone the repository:
git clone https://github.com/Solvve/ml_game_score_extraction.git
  • Install Tesseract-OCR >= 4.1.
  • Put fifa_score.traineddata file to %TESSDATA_PREFIX%.
  • Put fifa_scores.h5 file to %PROJECT_PATH%/data
  • Install all required dependencies:
pip install -r requirements.txt

Usage

python main.py

Application starts at http://localhost:8050/