Interface DataRequest

Maximum number of frames per chunk

Uploaded files. Must contain all files from job_file_mapping if job_file_mapping is not empty.

If not null, the files referenced by server_files will be retrieved from the cloud storage with the specified ID. The cloud storages applicable depend on the context. In the user sandbox, only the user sandbox cloud storages can be used. In an organization, only the organization cloud storages can be used.

Copy data from the server file share to CVAT during the task creation. This will create a copy of the data, making the server independent from the file share availability

A filename filter for cloud storage files listed in the manifest. Supports fnmatch wildcards. Read more: https://docs.python.org/3/library/fnmatch.html

Frame filter. The only supported syntax is: 'step=N'

Image quality to use during annotation

Represents a file-to-job mapping. Useful to specify a custom job configuration during task creation. This option is not compatible with most other job split-related options. Files in the jobs must not overlap or repeat. Job file mapping files must be a subset of the input files. If directories are specified in server_files, all files obtained by recursive search in the specified directories will be used as input files. In case of missing items in the input files, an error will be raised. Example: [ ["file1.jpg", "file2.jpg"], # job #1 files ["file3.png"], # job #2 files ["file4.jpg", "file5.png", "file6.bmp"], # job #3 files ]

Direct download URLs for files. Must contain all files from job_file_mapping if job_file_mapping is not empty.

Paths to files from a file share mounted on the server, or from a cloud storage. Must contain all files from job_file_mapping if job_file_mapping is not empty.

Paths to files and directories from a file share mounted on the server, or from a cloud storage that should be excluded from the directories specified in server_files. This option cannot be used together with filename_pattern. The server_files_exclude parameter cannot be used to exclude a part of dataset from an archive. Examples: Exclude all files from subfolder 'sub/sub_1/sub_2'and single file 'sub/image.jpg' from specified folder: server_files = ['sub/'], server_files_exclude = ['sub/sub_1/sub_2/', 'sub/image.jpg'] Exclude all cloud storage files with prefix 'sub' from the content of manifest file: server_files = ['manifest.jsonl'], server_files_exclude = ['sub/']

First frame index

Last frame index

Allows to specify file order for client_file uploads. Only valid with the "predefined" sorting method selected. To state that the input files are sent in the correct order, pass an empty list. If you want to send files in an arbitrary order and reorder them afterwards on the server, pass the list of file names in the required order.

Enable or disable task data chunk caching for the task. Read more: https://docs.cvat.ai/docs/manual/advanced/data_on_fly/

When true, video chunks will be represented as zip archives with decoded video frames. When false, video chunks are represented as video segments