QuanGo - is a system that automatically scores end game of the game of Go. Watch the demo here - DEMO

It consists of two modules: QNN - Neural Network written in Pytorch, it identifies a board from a picture

Score Calculation Backend - Implementation of static algorithm developed by Andrea Carta in Golang, original article can be accesed here

On this image you can see the architecture of the project

Here is example board and example result

Red color represents influence of white stones, while blue represents influence of black stones

QNN and Score Calculation Backend communicate through gRPC

message BoardRequest {
  repeated string board = 1; 
  int32 size = 2;            
  int32 black_prisoners = 3; 
  int32 white_prisoners = 4; 
  float komi = 5;            
}

message ScoreReply {
  float black_score = 1;
  float white_score = 2;
  string winner = 3;
}