LoL Replay Extractor

!! LoL Replay Extractor is still in development, any bugs can appear !!

`LoL Replay Extractor 🌟` is a program for gathering positional data and providing analytics from League of Legends videos. It can be used to automatically gather spatiotemporal data (player locations over time) from a series of Youtube/locally stored videos

✔ Any PRs are welcome, Watch nightly branch for recent updates!

✅What is pyLoL?

pyLoL is 'League of Legends' replays data extracting program.

✅What can pyLoL do?

[1] can save replay files(.rofl) automatically.

[2] can get the location of players every one second.

[3] can get the location of wards.

🚀Quickstart

• youtube_guide

Get started with pyLoL in nine steps:

1. First, clone this repository.

git clone https://github.com/league-of-legends-replay-extractor

2. Install Requirements

pip install -r requirements.txt

3. Make python project to Package

python setup.py develop

4. Directory settings

   In replay_scraper.ipynb, modify the path to suit your local environment

   • game_dir: League of Legends game directory.
   • replay_dir: League of Legends *.rofl replay directory.
   • dataset_dir: JSON replay files output directory.
   • replay_speed: League of Legends client replay speed multiplier.
   • scraper_dir: Directory of the scraper program.

rd.set_replays_dir(rd,folder_dir = r'C:\Users\username\Documents\League of Legends\Replays')  # replay download directory

rs.__init__(rs, game_dir = r'C:\Riot Games\League of Legends\Game',                          
            replay_dir = r'C:\Users\username\Documents\League of Legends\Replays',               
            dataset_dir = r'C:\Users\username\Documents\League of Legends\Dataset',              
            scraper_dir = r'C:\Users\username\Desktop\pyLoL\pyLoL\autoLeague\replays',
            replay_speed=40,
            region="KR")
ie.__init__(ie,dataset_dir=r'C:\Users\username\Desktop\pyLoL')

5. API KEY You can get API KEY from riot developer portal : https://developer.riotgames.com

dg.__init__(dg, api_key='RIOT_API_KEY' , count=20)

6. Gathering MatchIds filtered by {queue, tier, division, max_ids, patch_start_datetime}

   if you wanna download matchIds from 5000 MASTER I in SOLORANK users in specific patch,

dg.get_tier_matchIds(dg, queue='RANKED_SOLO_5x5', tier='MASTER', division='I' , max_ids=5000, patch_start_datetime='2023.10.26')

7. Save replay files for the match IDs obtained above

from tqdm import tqdm
import time

for matchId in tqdm(matchIds_challenger[:1000], 
                    desc = 'Gathering Replay_files(.rofl) from LoL Client... ', ## Print statement for progress at the front
                    ncols = 130, ## Adjust width for progress display
                    ascii = ' =', 
                    leave=True):
    
    try: #if this match id outdated, skip
        rd.download(rd, matchId)
        time.sleep(7)
    except:
        pass

8. Run replay => Save minimap capture video

• Option
  • All) no fog of war
  • Blue) fog of war in Blue team
  • Red) fog of war in Red team

for replay in tqdm(replays,
                    desc = 'Extracting Replay_Minimaps from LoL Client... ', ## Print statement for progress at the front
                    ncols = 200, ## Adjust width for progress display
                    ascii = ' =', 
                    leave=True
                    ):
    
    rs.run_client(rs,
                  replay_path = rf'{rs.get_replay_dir(rs)}\{replay}', 
                  gameId = replay.split('.')[0],
                  start=5*60 - 5, 
                  end=25*60 - 5, 
                  speed=10, 
                  paused=False, 
                  team="All")

9. Extract Realtime KDA, CS using OCR

from autoLeague.preprocess.ocr_center_window import OcrCenter as oc
oc.__init__(oc, project_folder_dir = r'C:\Users\username\Desktop\pyLoL')
oc.get_ocr(oc)

🎯Champion Tracking Model( Using Minimap Frame dataset )

• Champion Tracking (Roboflow)
• 

Then, You can use this tracking model ( Performance | mAP : 92.2% | precision : 91.3% | recall : 90.2% )

• Infer on Local and Hosted Images To install dependencies,

    pip install roboflow.
  

  Then, add the following the following code snippet to a Python script:

    from roboflow import Roboflow
    rf = Roboflow(api_key="API_KEY")
    project = rf.workspace().project("lolpago-multi-tracking-service")
    model = project.version(18).model
    
    # infer on a local image
    print(model.predict("your_image.jpg", confidence=40, overlap=30).json())
    
    # visualize your prediction
    # model.predict("your_image.jpg", confidence=40, overlap=30).save("prediction.jpg")
    
    # infer on an image hosted elsewhere
    # print(model.predict("URL_OF_YOUR_IMAGE", hosted=True, confidence=40, overlap=30).json())
  

🌠Give this Project a Star

This project is an open source project. If you like what you see, please give it a star on github.

☕Buy me a coffee

📝Citing this Paper

Win Prediction in MOBA Games Using Time-Series Location Data : Case Study in League of Legends (LoL)

  @article{김성윤2023시계열,
    title={시계열 위치 데이터를 활용한 MOBA 게임 승패 예측: 리그오브레전드 사례 분석},
    author={김성윤 and 이태홍 and 양희철},
    journal={한국통신학회 학술대회논문집},
    pages={1205--1206},
    year={2023}
  }

🖼️Update(Champ Detection Module)

• Champion Tracking (Not Roboflow, Just Using LoL DataDragon & Self-Augemted Dataset)
• Get champ portraits from riot developer's datadragon(https://ddragon.leagueoflegends.com/cdn/dragontail-14.20.1.tgz)
• Augment detection dataset from champ portraits & minimap
• Also add 'pings' and 'turrets' for better detection from 'Occlusion Problem' (eg. Back Pings Hides Champ Portraits @ minimap)
• No need to manually label champions on the minimap!
• In 14.20.1 patch, there are 168 champs

kimsy1106	AkiaCode	cye_dev
developing replay extractor	developing replay extractor	manage "read the docs"