cvat-ai/cvat v2.69.0 on GitHub

[SDK] TrackingFunction is now exported from cvat_sdk.auto_annotation (#10739)
Added configurable startup arguments and process counts for backend workers, including support for running webhook workers through rqworker-pool with a configurable number of processes (numWorkers)
numProcs is now passed as value, should not use additionalEnv anymore. NUMPROCS from values override additionalEnv value. (#10783)

SSRF mitigations are no longer applied to backing cloud storage requests (#10653)
[SDK] The x_organization parameter passed to individual API calls now takes precedence over the client-level organization_slug. Previously, organization_slug would silently override it, causing those calls to operate in the wrong organization context (#10665)
Updated backend worker liveness probes. Now it validates the expected worker count from chart values, so both regular workers and webhook worker that uses rqworker-pool can be checked correctly. (#10783)
Updated nuclio subchart version + dashboard version (#10792)
[Server API] Export and backup webhook payloads now report successful requests with the succeeded status instead of completed (#10734)
Allowed importing COCO and COCO Keypoints annotations without the iscrowd field. (cvat-ai/datumaro#138)

[SDK] Fixed Project.get_tasks(), Project.get_labels(), Task.get_jobs(), Task.get_labels(), Job.get_issues(), Job.get_labels(), and Issue.get_comments() methods returning an empty list when called on organization resources without an explicit organization context set on the client, or with a different organization context set (#10665)
Fixed incorrect responses being returned from the GET /api/quality/reports endpoint when the parent_id parameter refers to a report inaccessible by the current user (#10807)

Fixed a bug allowing a user to determine whether an inaccessible quality report is located in an organization they are not a member of or elsewhere (#10807)