REAL-Colon
收藏资源简介:
REAL-Colon数据集是由Cosmo Intelligent Medical Devices等机构联合创建,包含275万帧的高分辨率真实世界结肠镜检查视频,涵盖多个医疗中心。该数据集不仅包含35万个边界框标注,还附带了详细的临床数据和病理信息。创建过程中,专家胃肠病学家监督了标注过程,确保了数据的质量和准确性。REAL-Colon数据集的应用领域主要集中在提升人工智能在结肠镜检查中的辅助检测和诊断能力,旨在通过提供大规模、高质量和多样化的数据,推动相关算法和模型的研发与优化。
The REAL-Colon dataset was jointly created by Cosmo Intelligent Medical Devices and other institutions. It contains 2.75 million frames of high-resolution real-world colonoscopy videos sourced from multiple medical centers. This dataset not only includes 350,000 bounding box annotations, but also features detailed clinical and pathological information. Expert gastroenterologists supervised the annotation process throughout the dataset’s creation to ensure its quality and accuracy. The primary applications of the REAL-Colon dataset focus on improving the auxiliary detection and diagnostic capabilities of artificial intelligence in colonoscopy. It aims to advance the research, development and optimization of relevant algorithms and models by providing large-scale, high-quality and diverse medical data.




