Skip to content

Latest commit

 

History

History
47 lines (30 loc) · 1.42 KB

File metadata and controls

47 lines (30 loc) · 1.42 KB

Transition from Original Idea to something simplier.

Input: A screenshot from a minecraft world with a "digit" that is built.

The "digit" must be an image from minecraft (preferrably superflat world with a high contrast block) that is in the shape of the {letter, number, symbol} that you want decoded.

Input2: A world where you would have a 2x2 grid of blocks {9 distinct blocks varying in texture, and color} where you can have a mixture of different blocks on this "wall", in different positions (within the 2x2 grid).

  • Different types of blocks

Output2: A list of which blocks are used in the wall.

(1) Data Gathering

  • Input: Take a video of all of the pre-constructed {digits, symbols, number}
  • Output: Split the video into individual frames

(2) Encoding that data (machine interpretable)

  • Input: High Resolution Video Frame (1920x1080)
  • Ouput: Low Resolution image (100x100)

(3) Apply our ML Algorithms

  • (MVP): We can ask a pre-trained model (that is trained on digit recognition) what their prediction is
  • Create our own ML algorithm

(4) Decode into human readable

(5) Output

Model v1.0

Paramaters we can adjust to make it more "minecraftian"

  • Change the time of day
  • Change the weather
  • Change the blocks used to construct the image
  • Change the environment {not super-flat}

Output: A minecraft text output with the proper {letter, number, symbol} that you've created.