This is a fork over https://github.com/CarvellScott/OverwatchWatcher which does not seem to be developed anymore.
I'm not sure about you, but I have enough trouble memorizing all interactions
among the 18 types of Pokemon, let alone all of Overwatch's heroes. So I
made this script, intended to suggest counterpicks for me. The script is
completely client-side with no modification to the binary or manipulation
of network communications.
- CarvellScott
-
Install Python 3.5 (and pip, but it should come with python anyway)
-
Create a virtual environment
python -m venv ./venv
-
Install requirements
pip install -r requirements.txt
-
Set up OpenCV for Python 3. This might be more complicated, so look it up on Google :)
Either create screenshots ingame and tell the program their location, or use tab + a ingame to have the program take one.
- The script monitors the directory Overwatch saves its screenshots to, and opens the latest screenshot taken or takes a screenshot on it's own.
- An image recognition algorithm is used to guess what heroes are on the enemy team.
- Note: Image recognition only gives reliable results under certain restrictions. On Fire/Death and changing locations currently give uncertain results
- Using player-sourced data, it suggests heroes that have the biggest advantage against the enemy team.
- Once finished, it sounds a system alarm when done. Findings are recorded to the clipboard so you can paste them into in-game chat.
This was last tested on 07 Mei 2017.
At the moment, some modification of the code is needed to alter resolution of screenshots. Only works for 1440p right now