Based on image recognition technology, this paper studies the traditional music score conversion algorithm, aiming at realizing the automatic conversion from paper score to digital score. In this paper, a music score image recognition method based on convolutional neural network is proposed, and its effectiveness is verified by experiments. On this basis, the related technologies and problems of music score conversion are further discussed, including how to preprocess music score images, how to extract music score features and how to realize music score conversion algorithms. The research results show that the traditional music score conversion algorithm based on image recognition technology has high accuracy and robustness, and can effectively realize the digital conversion of music score.