SpectralWaste

SpectralWaste数据集是由萨拉戈萨大学创建的首个多模态数据集，专门用于废物分拣自动化。该数据集包含来自实际塑料废物分拣设施的同步高光谱和常规RGB图像，旨在识别和分离影响分拣效率的关键对象。数据集包含2059个标注实例，覆盖6个类别，适用于提高工业环境中复杂感知任务的性能。创建过程中，使用了两种同步的线扫描相机来捕捉RGB和高光谱数据，确保了数据的真实性和准确性。SpectralWaste数据集的应用领域主要集中在废物管理自动化，旨在通过提高分拣效率来减少环境负担，并保护工人健康。

SpectralWaste is the first multimodal dataset developed by the University of Zaragoza, specifically tailored for automated waste sorting. It contains synchronized hyperspectral and conventional RGB images collected from real plastic waste sorting facilities, with the purpose of identifying and isolating key objects that impede sorting efficiency. The dataset includes 2059 annotated instances across 6 categories, and is intended to enhance the performance of complex perception tasks in industrial environments. During its development, two synchronized line-scan cameras were employed to capture RGB and hyperspectral data, ensuring the authenticity and accuracy of the collected data. The primary application domains of the SpectralWaste dataset focus on waste management automation, aiming to improve sorting efficiency, reduce environmental burdens, and protect workers' health.