Currently under development, this project is a work in progress. Please check back later for updates.
The project aims to detect dart scores by using a yolo v11 keypose model on a single camera image to predict the score of the thrown darts. 4 Keypoints on the dartboard, the darts tip and flight are detected.
After detecting the keypoints, the 4 dartboard keypoints are used to apply a perspective transform to warp the perspective in a top down view. With the warped points of the dart tips the score of the dart can easily be calculated by their distance to the center and their angle.
To see how to train your own model and how to prepare and create a dataset, refer to Train Instructions