Task recipes

Three recipes cover the task lifecycle: task_create_from_cloud.py creates one task from object keys already in a registered bucket, tasks_bulk_from_cloud.py creates a whole batch of tasks in a project from that same bucket, and task_inspect_and_export.py inspects an existing task, exports its dataset locally, and reports analytics from its event log.

Create a task from cloud object keys

Creates a task from images that already live in a registered bucket.

python task_create_from_cloud.py --host 'https://app.cvat.ai' --token '<your token>' \
    --cloud-storage-id 7 --cloud-keys 'images/0001.jpg' 'images/0002.jpg' \
    --labels car person

The script

{{< include-code "assets/sdk-examples/task_create_from_cloud.py" >}}

Bulk-create tasks in a project from a bucket

Creates several tasks in one call, all inside the same project, each reading its data from a registered cloud storage. Two ways to spell a task's data, repeatable and mixable: --task KEY[,KEY,...] lists explicit object keys (a single key makes a video/single-image task; multiple keys make an image task whose frames are those keys in order), and --task-pattern PATTERN makes one task from every bucket file matching a fnmatch wildcard (e.g. 'batch_a/*.jpg'), resolved from the bucket's manifest instead of listing every key by hand. Because every task belongs to the project, they share its label schema — no --labels here.

# three video tasks in project 42
python tasks_bulk_from_cloud.py --host 'https://app.cvat.ai' --token '<your token>' \
    --cloud-storage-id 7 --project-id 42 \
    --task 'videos/clip_01.mp4' --task 'videos/clip_02.mp4' --task 'videos/clip_03.mp4'

# two image-batch tasks in project 42
python tasks_bulk_from_cloud.py --host 'https://app.cvat.ai' --token '<your token>' \
    --cloud-storage-id 7 --project-id 42 \
    --task 'batch_a/img_1.jpg,batch_a/img_2.jpg' \
    --task 'batch_b/img_1.jpg,batch_b/img_2.jpg'

# the same two batches, without listing every key: one task per wildcard match
python tasks_bulk_from_cloud.py --host 'https://app.cvat.ai' --token '<your token>' \
    --cloud-storage-id 7 --project-id 42 --manifest manifest.jsonl \
    --task-pattern 'batch_a/*.jpg' --task-pattern 'batch_b/*.jpg'

The script

{{< include-code "assets/sdk-examples/tasks_bulk_from_cloud.py" >}}

Inspect a task and export its dataset

Prints a summary of an existing task (labels, jobs, frames), exports its dataset to a local zip, then exports the task's event log and reports two analytics computed from it: how many people are currently assigned to a job, and how many jobs were rejected in review and sent back for rework.

python task_inspect_and_export.py --host 'https://app.cvat.ai' --token '<your token>' \
    --task-id 42 --export-format 'COCO 1.0'

The script

{{< include-code "assets/sdk-examples/task_inspect_and_export.py" >}}

Other SDK options:

Notes:

• To add a task to a project, pass project_id in TaskWriteRequest and do not pass labels — the task inherits the project's label schema.
• Both cloud recipes use ResourceType.SHARE, so the images are read from the bucket rather than uploaded from your machine.
• include_images=False exports annotations only and is much smaller.
• Pass a valid format_name from the server's exporter list, e.g. "COCO 1.0" or "CVAT for images 1.1". An unknown format name is rejected by the recipe.
• Full recipes: task_create_from_cloud.py, tasks_bulk_from_cloud.py, task_inspect_and_export.py.