Computer Vision Annotation Tool (CVAT) Files

Added

• Added registration survey fields to user profiles and made the primary role editable on the profile page (https://github.com/cvat-ai/cvat/pull/11216)

Changed

• Multi-selection now uses fixed Shift-drag and Ctrl/Cmd-click modifiers instead of configurable shortcuts, eliminating false conflict warnings involving these modifiers (https://github.com/cvat-ai/cvat/pull/11281)

• Improved annotation workspace performance for jobs with many objects (https://github.com/cvat-ai/cvat/pull/11220)

Fixed

• Improve performance while using the brush tool for larger segments (https://github.com/cvat-ai/cvat/pull/11119)

• Model loading now waits for organization initialization, preventing functions from switching organization context between pages. (https://github.com/cvat-ai/cvat/pull/11266)

• Copying and pasting skeleton tracks with attributes could fail or lose keypoint attribute values (https://github.com/cvat-ai/cvat/pull/11279)

• Keyboard shortcuts preventing selected text from being copied and interfering with text editing. (https://github.com/cvat-ai/cvat/pull/11280)

• Fixed intermittent generic errors when retrieving chunks so the API reliably reports the underlying failure (https://github.com/cvat-ai/cvat/pull/11282)

• Links in quality control card tooltips could be difficult to click because the tooltip closed too quickly (https://github.com/cvat-ai/cvat/pull/11284)

Security

• Fixed disclosure of user/task names and IDs via the browsable API (https://github.com/cvat-ai/cvat/security/advisories/GHSA-pv9q-68mq-qwpm)

• Fixed multiple path traversal vulnerabilities (https://github.com/cvat-ai/cvat/security/advisories/GHSA-qpmj-jxq7-67fg)