/mcbc_eval

Client integration for Minecraft biome classification

Primary LanguageJavaOtherNOASSERTION

MCBC_Eval

Minecraft Fabric 1.19.3 mod featuring live biome classification using the MCBC Convolutional Neural Network trained on data created by the MCBC Data Generator. This project consists of two parts:

  • Java Clientside Fabric Mod (Frontend): Generates the screenshots and outputs the evaluation results in game.
  • Python CNN (Backend) Hosts the pretrained CNN and evaluates the generated images.

Both ends communicate using the XML-RPC Protocol, but are required to run on the same device.

⚠️Please Note: This repository does not include the pretrained weights of the CNN, only its architecture.

It is necessary to download the pretrained weights (252 MB) prior to running the Python CNN backend. Save the file under the subpath /cnn/mcbc_weights.h5 of this workspace.

Python Dependencies

  • Python 3
  • keras
  • tensorflow
  • scikit-image

Team

Sven Schreiber (https://github.com/svenschreiber)
Lasse Huber-Saffer (https://github.com/OfficialLahusa)
Nico Hädicke (https://github.com/Reshxram)
Lena Kloock (https://github.com/LenaKloock)