Epic 7 Stats

This website aims to help players climb RTA more effectively by providing provides information about popular heroes and their builds using data collected from content creators' videos.

Datamine RTA Matches

A python program was created to analyze streamers' videos, to identify heroes selected in draft.

A summary of how the program works:

  1. Prematch: read text on the upper left corner to identify game mode
  2. Match start: use OCR to read players info such as league, server ...
  3. Draft Phase:
    • get prebanned heroes, first pick player
    • detect stage of the draft, when stage changes detect contours in the frame
    • remove contours doesn't meet requirements(size, location),
    • compare each contour with all heroes' icon to find the best match.
  4. Battle phase: wait for Confirm button on lower right then classify win/loss.



Working on improving data collection and visualization.

Adding more interesting Epic7 statistics.

July 2021

Redesigned website


  • Web Framework: Python Flask, Gunicorn
  • Database: MongoDB
  • Web server: Nginx
  • Host: AWS EC2


  • UI, HTML, CSS: modified Pixel theme from themesberg.com
  • Chart: Apex Chart JS

Feb 2021

First Dashboard Released

Built a dashboard using Python Streamlit framework and deployed on Heroku.

Jan 2021

Started collecting RTA data

Using Python OpenCV for hero recognition and Tesseracts OCR for builds.