ALFI dataset (final)

Data is divided into three folders: 1) Folder "Data&Annotations" includes images and annotations for all the sequences. 1.1) For sequences MI01 through MI08 (including annotations for both Task 1 and Task 2), the content is, e.g., ├── Data&Annotations │ ├── MI01 │ │ ├── Images │ │ │ ├── I_MI01_0001.png ... │ │ │ └── I_MI01_0069.png │ │ ├── MI01_DTLTruth.csv │ │ ├── MI01_PhenoTruth.csv │ │ └── Masks │ │ ├── M_MI01_0001.png ... │ │ └── M_MI01_0069.png For each j, I_MI01_j.png is the j-th image of video MI01 and M_MI01_j.png is the corresponding ground truth segmentation mask. The MI01_DTLTruth.csv file includes annotation information consisting of a) Number of sequence image (ImNo), b) Cell ID (ID), c) Cell class (Class, either Interphase or Mitosis), d) Bounding box, specified by its upper left corner (xmin, ymin) and its dimensions (width, height), and e) Parent cell ID (Parent). The MI01_PhenoTruth.csv file includes annotation information consisting of a) Number of sequence image (ImNo), b) Cell ID (ID), c) Cell class (Class, including EarlyMitosis, LateMitosis, CellDeath, and Multipolar), and d) Bounding box, specified by its upper left corner (xmin, ymin) and its dimensions (width, height). 1.2) For sequences CD01 through CD09 and TP01 through TP12 (including only annotations for Task 2), the content is similar, but there is no "Masks" subdirectory nor "*PhenoTruth.csv" file. 2) Folder "UseExamples" includes two usage examples and two directories including their output, in the form of Matlab/Python scripts. ├── UseExamples │ ├── UseExample1.m │ ├── UseExample1.py │ ├── UseExample1.py.readme.txt │ ├── UseExample1Out │ │ ├── TP06BBannots │ │ │ ├── TP06_py_WithBB_0001.png ... │ │ │ └── TP06_py_WithBB_0016.png │ │ │ ├── TP06_WithBB_0001.png ... │ │ │ └── TP06_WithBB_0016.png │ ├── UseExample2.m │ ├── UseExample2Out │ │ ├── MI01Lineage.png │ │ ├── MI01LineageCrops.png │ │ ├── MI02Lineage.png │ │ ├── MI02LineageCrops.png │ │ ├── MI03Lineage.png │ │ ├── MI03LineageCrops.png │ │ ├── MI04Lineage.png │ │ ├── MI04LineageCrops.png │ │ ├── MI05Lineage.png │ │ ├── MI05LineageCrops.png │ │ ├── MI06Lineage.png │ │ ├── MI06LineageCrops.png │ │ ├── MI07Lineage.png │ │ ├── MI07Lineage(Mitoses).png │ │ ├── MI07LineageCrops.png │ │ ├── MI08Lineage.png │ │ ├── MI08Lineage(Mitoses).png │ │ ├── MI08LineageCrops.png 2.1) UseExample1.m is a Matlab script for plotting the bounding box annotations over the original sequence images; the output is included in folder UseExample1Out/TP06BBannots, that contains images TP06_WithBB_0001.png, ..., TP06_WithBB_0016.png. UseExample1.py is the analogous examples written in Python and UseExample1.py.readme.txt is the related readme file; the output is included in folder UseExample1Out/TP06BBannots, that contains images TP06_py_WithBB_0001.png, ..., TP06_py_WithBB_0016.png. 2.2) UseExample2.m is a Matlab script for plotting the cells lineage for one of the sequences MI01-MI08; the output of this example is the lineage representation and image crops of mitotic cells given in the PNG images included in folder UseExample2Out. For the user convenience, we provide the output images for all MI* sequences, without need to run it. Please, observe that, for the very crowded sequences MI07 and MI08, we also also provide the lineage representation reduced only to Mitoses (e.g., MI07Lineage(Mitoses).png), obtained setting the parameter OnlyMitoses = 1; (line 49 of the script). 3) Folder "Videos" includes the 29 videos stored in ND2 format. └── Videos ├── CD01.nd2 ... └── TP10.nd2

数据集共分为三个文件夹：
1. "数据与标注（Data&Annotations）"文件夹：包含所有序列的图像与标注文件。
1.1 针对MI01至MI08序列（覆盖任务1与任务2的标注需求），其目录结构示例如下：
├── "Data&Annotations"
│ ├── MI01
│ │ ├── Images
│ │ │ ├── I_MI01_0001.png …
│ │ │ └── I_MI01_0069.png
│ │ ├── MI01_DTLTruth.csv
│ │ ├── MI01_PhenoTruth.csv
│ │ └── Masks
│ │ ├── M_MI01_0001.png …
│ │ └── M_MI01_0069.png
其中对于每个序号`j`，`I_MI01_j.png`为视频MI01的第`j`帧图像，`M_MI01_j.png`为对应的**真值分割掩码（ground truth segmentation mask）**。`MI01_DTLTruth.csv`文件包含以下五类标注信息：a) 序列图像编号（ImNo）、b) 细胞ID（ID）、c) 细胞类别（Class，分为间期（Interphase）与有丝分裂期（Mitosis）两类）、d) 由左上角坐标`(xmin, ymin)`与尺寸`(width, height)`定义的**边界框（Bounding box）**、e) 母细胞ID（Parent）。`MI01_PhenoTruth.csv`文件包含以下四类标注信息：a) 序列图像编号（ImNo）、b) 细胞ID（ID）、c) 细胞类别（Class，包含早有丝分裂（EarlyMitosis）、晚有丝分裂（LateMitosis）、细胞死亡（CellDeath）与多极分裂（Multipolar））、d) 由左上角坐标`(xmin, ymin)`与尺寸`(width, height)`定义的边界框。
1.2 针对CD01至CD09以及TP01至TP12序列（仅包含任务2的标注），目录结构与上述示例一致，但无`Masks`子目录，也不存在`*PhenoTruth.csv`文件。
2. "使用示例（UseExamples）"文件夹：包含2个使用示例与2个存储其输出结果的目录，示例代码格式为Matlab/Python脚本。
其目录结构示例如下：
├── "UseExamples"
│ ├── UseExample1.m
│ ├── UseExample1.py
│ ├── UseExample1.py.readme.txt
│ ├── UseExample1Out
│ │ ├── TP06BBannots
│ │ │ ├── TP06_py_WithBB_0001.png …
│ │ │ └── TP06_py_WithBB_0016.png
│ │ │ ├── TP06_WithBB_0001.png …
│ │ │ └── TP06_WithBB_0016.png
│ ├── UseExample2.m
│ ├── UseExample2Out
│ │ ├── MI01Lineage.png
│ │ ├── MI01LineageCrops.png
│ │ ├── MI02Lineage.png
│ │ ├── MI02LineageCrops.png
│ │ ├── MI03Lineage.png
│ │ ├── MI03LineageCrops.png
│ │ ├── MI04Lineage.png
│ │ ├── MI04LineageCrops.png
│ │ ├── MI05Lineage.png
│ │ ├── MI05LineageCrops.png
│ │ ├── MI06Lineage.png
│ │ ├── MI06LineageCrops.png
│ │ ├── MI07Lineage.png
│ │ ├── MI07Lineage(Mitoses).png
│ │ ├── MI07LineageCrops.png
│ │ ├── MI08Lineage.png
│ │ ├── MI08Lineage(Mitoses).png
│ │ ├── MI08LineageCrops.png
2.1 `UseExample1.m`为Matlab脚本，用于在原始序列图像上绘制边界框标注，其输出结果存放于`UseExample1Out/TP06BBannots`文件夹中，包含`TP06_WithBB_0001.png`至`TP06_WithBB_0016.png`等图像。`UseExample1.py`为等效的Python版本示例，`UseExample1.py.readme.txt`为相关说明文档，其输出结果同样存放于`UseExample1Out/TP06BBannots`文件夹中，包含`TP06_py_WithBB_0001.png`至`TP06_py_WithBB_0016.png`等图像。
2.2 `UseExample2.m`为Matlab脚本，用于绘制MI01至MI08中任一序列的**细胞谱系（cell lineage）**，该示例的输出结果为存放于`UseExample2Out`文件夹中的PNG格式图像，包含细胞谱系表征与有丝分裂细胞的裁剪图像。为方便用户使用，我们已提供所有MI*序列的输出图像，无需用户自行运行脚本。请注意，对于样本密度较高的序列MI07与MI08，我们还额外提供了仅保留有丝分裂细胞的谱系表征（例如`MI07Lineage(Mitoses).png`），该结果通过将脚本第49行的`OnlyMitoses`参数设为1得到。
3. "视频（Videos）"文件夹：包含29个以ND2格式存储的视频文件。
其目录结构示例如下：
└── "Videos"
├── CD01.nd2
…
└── TP10.nd2