FoodSeg103

FoodSeg103是由新加坡管理大学构建的大型食品图像分割数据集，包含9490张图像，涵盖154种食材类别。该数据集通过精细标注，每张图像平均包含6种食材标签及像素级掩码，旨在支持食品营养和热量估算等健康相关应用的研究。创建过程中，研究团队进行了严格的数据筛选和标签迭代优化，确保了数据集的高质量。FoodSeg103的应用领域主要集中在食品图像的细粒度理解和分析，为食品计算领域提供了重要的基准数据集。

FoodSeg103 is a large-scale food image segmentation dataset constructed by Singapore Management University. It contains 9,490 images covering 154 ingredient categories. With fine-grained annotations, each image has an average of 6 ingredient labels paired with pixel-level masks. This dataset aims to support research on health-related applications such as food nutrition and calorie estimation. During its development, the research team conducted strict data screening and iterative label optimization to ensure its high quality. The main application areas of FoodSeg103 focus on fine-grained understanding and analysis of food images, serving as an important benchmark dataset for the field of food computing.