ChessBoardDetector

Two-model YOLOv11 pipeline that converts a chessboard photograph into a FEN string.

Models

File	Architecture	Task
board_detector.pt	YOLOv11n	Locate the chessboard bounding box
piece_detector.pt	YOLOv11s	Identify all 12 piece types on a 512×512 rectified board

Usage

from src.pipeline import run_pipeline
import cv2

image = cv2.imread("photo.jpg")
fen, annotated = run_pipeline(
    image,
    board_model_path="board_detector.pt",
    piece_model_path="piece_detector.pt",
)
print(fen)

Source

GitHub — honi05/ChessBoardDetector