Awesome

BoardGame-Detection

Client Server Architecture

The client will send an image to the API server.
The server will response with the found pieces on the board

Client

There are two main clients:

Raspberry Pi Client PiClient
StreamClient

Raspberry Pi Client

The PiClient must sit on the Raspberry Pi. It will connect to the Pi Camera with the picamera library. It will take an image from the Pi Camera and sends it to the API Server.

This Raspberrry Pi Client is using Pillow instead of opencv-python to remove the build process of OpenCV.

Make sure you installed the correct packages for Pillow on the Pi with:

sudo apt-get install python-pil

Python dependencies

numpy
pillow
aiohttp
picamera

from client.src.piclient import PiClient

# Setup the PiClient
client = PiClient(
    url='http://127.0.0.1',
    token='a_random_token_please'
)

# Ping the API server to check its uptime
#   This will return either True or False
print(client.ping())

# Send an image to detect the figures
#   This will return two variables
#       1. the list of pieces for the game logic
#       2. debug infos, if client.detect(debug=True)
pieces, _ = client.detect()
print(pieces)

StreamClient

The StreamClient can sit anywhere. It can access a video stream of another device. It will take the latest frame of the stream and sends it to the API Server.

Python dependencies

numpy
pillow
aiohttp
opencv-python
imutils

from client.src.streamclient import StreamClient

# Setup the StreamClient
client = StreamClient(
    url='http://127.0.0.1',
    token='a_random_token_please',
    stream_url='http://127.0.0.1:8000/stream.mjpg'
)

# Ping the API server to check its uptime
#   This will return either True or False
print(client.ping())

# Send an image to detect the figures
#   This will return two variables
#       1. the list of pieces for the game logic
#       2. debug infos, if client.detect(debug=True)
pieces, _ = client.detect()
print(pieces)

Server

Python dependencies

opencv-python
numpy
fastapi
uvicorn