Skip to content

A Ball Blast Game that uses the transfer learning and KNN algorithm as a backbone to train the model in the browser and use that trained model to control the movements of canon(ie. move left, move right, stay still).

Notifications You must be signed in to change notification settings

Asto7/ball-blast-ai

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ball-blast-ai (Link)

A game where a player has to hit balls falling from the sky by using trusty cannon (Gun). Player has to stop the balls from falling before they hit the cannon. Once the strength of a ball reaches to 0 from an initial level it again replicates itself with half the amount of initial strength and it goes that way. It raises the difficult level each time the balls count drops to zero.

Example Demo


Link for Game Alone without visual controls feature.

Visual Control Feature: User can train the canon movements(like move left, move right, stay still) on the browser itself with the help of mobilenet model, knn algorithm to classify and webcam to capture images for training the model.

Setup

npm install

Starting server

npm start

How to Train the Model In Browser

  • Long Press On the Class Label Button for which you want to train the model, meanwhile webcam will capture the images(Current position) and train it under that class label :)

TODO

  1. Add touch controls for mobile devices.
  2. Generate random balls.
  3. Better UI.

About

A Ball Blast Game that uses the transfer learning and KNN algorithm as a backbone to train the model in the browser and use that trained model to control the movements of canon(ie. move left, move right, stay still).

Topics

Resources

Stars

Watchers

Forks

Languages