未明确提及

本研究针对生物医学研究中的有丝分裂检测任务，提出了一种基于部分标注序列的检测方法。该方法通过帧顺序翻转和alpha混合粘贴技术，从部分标注数据生成完全标注的数据集，用于训练有丝分裂检测模型。数据集内容涉及多种荧光显微镜图像序列，包括HeLa、ES、ES-D和Fib等细胞类型，用于评估和验证方法的有效性。创建过程中，研究者利用部分标注信息，通过创新的图像处理技术生成训练数据，解决了传统完全标注数据集制作成本高和耗时的问题。该数据集主要应用于细胞生物学研究、医学诊断和药物开发等领域，旨在提高有丝分裂检测的准确性和效率。

This study proposes a detection method based on partially annotated sequences for the mitosis detection task in biomedical research. This method generates fully annotated datasets from partially annotated data using frame sequence flipping and alpha-blending pasting techniques, which are employed to train mitosis detection models. The datasets cover multiple fluorescence microscopy image sequences of various cell types including HeLa, ES, ES-D, Fib and others, and are utilized to evaluate and validate the effectiveness of the proposed method. During the dataset construction process, researchers leverage partially annotated information and innovative image processing technologies to generate training data, resolving the problems of high production cost and time-intensive workloads associated with traditional fully annotated datasets. This dataset is primarily applied in fields such as cell biology research, medical diagnosis and drug development, with the goal of enhancing the accuracy and efficiency of mitosis detection.