Training dataset for HeLa "Kyoto" cells under the scope

The Patches folder contains the training patches with csv files of cells 2 steps before the division and 2 steps in time after the division for 5 channels of the Kyoto Hela cells, the Oneat folder contains the json file of the cell categories and the map of XYTHW locations and its corresponding integer labels. The dataset is intended to be used for training oneat networks for these cells in different imaging modalities. Training patches with labels for 5 channels of hela cells: trainX, trainY = 64, 64, tminus = 2, tplus = 2, whole image was normalized using percentile based normalization pre patch making process. Patches made without segmentation image as input so height and width = 1 in yolo training labels. Made by oneat software, pip install oneat for training of such datasets. Original data published by Romain Guiet at https://zenodo.org/record/6139958#.YmAh1NpBxhG图像块文件夹（Patches folder）包含针对京都海拉细胞（Kyoto Hela cells）5个成像通道的训练图像块，配套的CSV文件记录了细胞分裂前2个时序步长、分裂后2个时序步长的相关数据。Oneat文件夹则包含细胞类别JSON文件、XYTHW位置映射表及其对应的整数标签。本数据集旨在针对不同成像模态下的此类细胞，用于训练Oneat网络。针对海拉细胞5个通道的带标签训练图像块参数如下：训练图像尺寸trainX、trainY均为64×64，tminus=2、tplus=2；在生成图像块前，已通过基于百分位数的归一化方法对整幅图像完成归一化处理。本次生成的图像块未以分割图像作为输入，因此在YOLO训练标签中，其高度与宽度均为1。本数据集由Oneat软件生成，若需训练此类数据集，可执行`pip install oneat`命令安装相关工具。原始数据由Romain Guiet发布于：https://zenodo.org/record/6139958#.YmAh1NpBxhG