The goal is to create a toolbox for League of Legends players to improve their skills in rank games. We build our tool using both machine learning and our own game understanding, this could help us handle the problems might appear in our ML models, or even design our own model to handle the data we collected.
It is not a guidance tool or an introduction of this game, the target users are those players who want to win more in ranks but have no time to watch pro videos and study the counter picks or item builds. The assistant also helps users getting the most useful information about the minimap, without kepping eyes on it.
victory never this easy
Ban/Pick Assistant
We build this based on screen capture and image detection; we trained a model using the profile image of 148 champions, this model is been used to detect the banned champion identity, use them as the input of our model, give a suggestion for our user to pick.
In-game Item build Advice
The detection result (including enemy builds and your current build) once you clicked “Tab” to check the score board. When you recalled back to your basement, and click the shop, our build suggestion will show as a litter black suggestion box on the top right corner, you can click the items in it to search in the shop, and will never blind anything on the map.
Minimap Warning
The assistant keeps tracking enemy’s icons on minimap, and got screen shot each second, then it will calculate the distance between each enemy champion and your line. It also keep tracking enemy’s location and calculate the distance between them and Baron/Dragon, if they keep getting close to these 2 positions for 2 to 3 seconds in last 5 seconds, and their location 5 seconds ago is not too far from these 2 positions, you will get an alert.