League of Legends Gamve Video OCR

Analyzing League of Legends game video with GCP OCR

Introduction

This repository contains code for Optical Character Recognition on League of Legends game video. The purpose of this project is as follows:

• Extract meaningful text values such as gold, level, KDA from League of Legends game video using GCP OCR API
• Develop various scenarios on what analysis can be done based on the extracted values.

Dependencies

The code is compatible with python3 and the following dependencies are needed to run the program.

• pandas == 1.1.5
• google-cloud-videointelligence == 1.16.0
• opencv-python == 4.2.0.34
• google-cloud-vision == 2.0.0
• pillow == 8.0.1
• tqdm == 4.51.0
• pytube == 10.0.0

Please use the command below to install all the dependencies.
!pip install -r requirements.txt --use-feature=2020-resolver

Installation

1. Clone the repository: https://github.com/nwojke/deep_sort.git

2. Create credential.json file from GCP If you are a member of the project, please create the credential file from OPGG service account

   you can create the file from GCP - IAM & ADMIN - Service Accounts - Actions - Create Key
   

   If you are not a member of the project, please create your credential file according to the instruction of Google Cloud Vision API
   

   After your done with creating credential file, please loacte the file in the same path as the run.py file

Modify Path

Please modify the following path of the constnats.py file that matches your OS.

• VIDEO_PATH
• IMAGE_PATH
• CSV_PATH

Download Game Video from Youtube

1. modify youtube_list.txt file and add the url you want to download
   
2. Run the program by using the following command and press 1
   python run.py

Generating OCR Tracker

1. Run the program by using the following command and press 2
   python run.py

Highlevel overview of source files

In package conver_ocr.py is the main tracking code:

• bitwise_operation.py: bitwise operation for mask to increase OCR performance
• constants.py: Path and coordinates of the video
• preprocess.py: A module to preprocess OCR text results
• video_download.py: Pytube module to download youtube video using url
• run.py: A module to run the tracker