RealmParser

A python module using OCR to processing first-person League of Legends videos (ideally for Tournament Realm games)

Install

Download repository and use run.py

Python Package Dependencies:

• opencv-python
• easyocr
• numpy
• pprint
• pandas (not necessary, but highly recommended)

Usage

The most important thing that you'll need are:

• a video to parse
• the game.cfg file of the player whose video is being parsed

import realmparser
import pandas as pd

video_name = "mid"
mid_realm = realmparser.VodParse(
    realm_path=f"{video_name}.mp4",
    input_region="scoreboard",
    lol_cfg_path="game.cfg",
    role="MID",
    des_time_interval=10,

)


df = pd.DataFrame.from_dict(mid_realm.output)
df.to_pickle("midrealm_data.pkl")
df.to_csv("midrealm_data.csv")

Time Required Benchmarks:

TODO

1. New regions and subregions:

- Map: TurretPlates, TurretStatus
- Scoreboard: RedTeamKills, BlueTeamKills
- ChampHUD: ChampionPlayed, GoldInHand, TotalGold, ItemsBuilt

2. Restructure output to conform with the loldto:

game: dict
├── sources: dict
├── teams: dict
|   ├── uniqueIdentifiers: dict
│   ├── bans: list
│   ├── monstersKills: list
│   ├── buildingsKills: list
│   └── players: list
│       ├── uniqueIdentifiers: dict
│       ├── endOfGameStats: dict
│       │   └── items: list
│       ├── summonersSpells: list
│       ├── runes: list
│       ├── snapshots: list
│       ├── itemsEvents: list
│       ├── wardsEvents: list
│       └── skillsLevelUpEvents: list
├── kills: list
└── picksBans: list

3. Determine whether list of images can be pulled from a VOD to use list comprehension instead of a for loop.