DeepScores

DeepScores 包含高质量的乐谱图像，分为 300,000 张书面乐谱，其中包含不同形状和大小的符号。用于推进小物体识别的最新技术，并将物体识别问题置于场景理解的背景下。

DeepScores is a dataset containing high-quality sheet music images, which are grouped into 300,000 sheet music pieces with symbols of varying shapes and sizes. It is developed to advance state-of-the-art technologies for small object recognition, and situates the object recognition problem within the context of scene understanding.

搜集汇总

背景与挑战

背景概述

DeepScores是一个高质量的乐谱图像数据集，包含30万张书面乐谱图像，旨在推动小物体识别和场景理解技术的研究。该数据集由达勒莫勒人工智能研究所于2018年发布。