Knowledge Integrated Dataset (KID)

知识集成数据集(KID)是由佐治亚理工学院的研究团队创建，旨在通过整合多个车辆识别数据集，解决车辆识别领域的数据集不兼容问题。KID包含约50万条数据，涵盖颜色、类型和制造商等多种车辆特征，数据来源于CompCars、BoxCars116K、Cars196等多个现有数据集。创建过程中，研究团队采用了一种名为ATEAM的专家标注团队方法，通过跨数据集标注和知识转移，实现了不同数据集间的特征集成。KID的应用领域广泛，包括车辆型号识别、车辆重识别、跟踪等，旨在提高现有模型的准确性和鲁棒性。

Knowledge Integration Dataset (KID) was developed by a research team from the Georgia Institute of Technology, aiming to address the dataset incompatibility issue in the vehicle recognition domain by integrating multiple vehicle recognition datasets. KID contains approximately 500,000 data entries, covering various vehicle attributes such as color, type and manufacturer, with data sourced from multiple existing datasets including CompCars, BoxCars116K, Cars196 and others. During the development process, the research team adopted an expert annotation team methodology named ATEAM, achieving feature integration across different datasets through cross-dataset annotation and knowledge transfer. KID has a broad range of application fields, including vehicle model recognition, vehicle re-identification, tracking and more, with the objective of enhancing the accuracy and robustness of existing models.