Export segmentation masks

You can export to your disk the pixel level segmentation masks of your segmentation projects.

Dependencies

This export requires additional dependencies that you can install with pip install isahitlab[image]

Code Reference

See the code reference for further details.

Compatibility

Export of masks is available for:

• Project type Segmentation (iat-segmentation)
• Project type Data processing (form) with input type:
  • Segmentation (tool-iat-segmentation)

For Data processing project

The SDK will try to detect the input compatible with the export. If more than one input is compatible, you must provide the input_id parameter to select the input to export.

Output

Folder containing .png mask

lab_mask_`project_id`_`datetime`/
├── image_07750.jpg.png
├── image_07751.jpg.png
└── image_07752.jpg.png

Usage

from isahitlab.client import IsahitLab

lab = IsahitLab()

lab.export_tasks(project_id="<project_id>", format="mask")

You can filter the tasks with the same parameters than you can use to get tasks.

from isahitlab.client import IsahitLab

lab = IsahitLab()

lab.export_tasks(project_id="<project_id>", 
                 format="mask", 
                 batch_id_in=["<batch_id>"], 
                 status_in=["complete", "reviewed"], 
                 updated_at_gte="2024-12-25 00:00:00"
                 )

Use output_folder and / or output_filename to choose where to save the results.

from isahitlab.client import IsahitLab

lab = IsahitLab()

lab.export_tasks(project_id="<project_id>", 
                 format="mask", 
                 output_folder="./output",
                 output_filename="my-export"
                 )

Output name

If you set the output_filename parameter, it will be used as folder name

Directory tree

The SDK will automatically create the folder tree if you set an output_folder like output/my_outputs/<project_id>

Options

replace_extension

Type: boolean

For each mask, the file name will be the name of the task with .png. If, for example, the name of your task is my-task-01.jpg, set this option to True to save your mask as my-task-01.png instead of my-task-01.jpg.png.

from isahitlab.client import IsahitLab

lab = IsahitLab()

lab.export_tasks(project_id="<project_id>", 
                 format="mask", 
                 output_folder="./output",
                 output_filename="my-export",
                 options={
                    "replace_extension" : True
                 }
                )

semantic_mapping

Type: Dict[str, (int,int,int)]

By defaut, each labeled pixel in your mask follows this RGB convention:

R	G	B
instance_id	instance_id % 255	Label bChannel (in your project configuration)

You can transform you mask to get custom semantic colored mask by passing a Dict where the keys are the label id configured on your project and the values are the RGB triplet to set.

from isahitlab.client import IsahitLab

lab = IsahitLab()

lab.export_tasks(project_id="<project_id>", 
                 format="mask", 
                 output_folder="./output",
                 output_filename="my-export",
                 options={
                    "semantic_mapping" : {
                        "Test 1":  (128, 64, 128),
                        "Test 2": (70, 70, 70),
                        "Test 3": (220, 220, 0)
                     }
                 }
                )