cvat

[0.6.0] - 2020-03-15

Added

• Server only support for projects. Extend REST API v1 (/api/v1/projects*)
• Ability to get basic information about users without admin permissions (#750)
• Changed REST API: removed PUT and added DELETE methods for /api/v1/users/ID
• Mask-RCNN Auto Annotation Script in OpenVINO format
• Yolo Auto Annotation Script
• Auto segmentation using Mask_RCNN component (Keras+Tensorflow Mask R-CNN Segmentation)
• REST API to export an annotation task (images + annotations)
• Datumaro - a framework to build, analyze, debug and visualize datasets
• Text Detection Auto Annotation Script in OpenVINO format for version 4
• Added in OpenVINO Semantic Segmentation for roads
• Ability to visualize labels when using Auto Annotation runner
• MOT CSV format support (#830)
• LabelMe format support (#844)
• Segmentation MASK format import (as polygons) (#1163)
• Git repositories can be specified with IPv4 address (#827)

Changed

• page_size parameter for all REST API methods
• React & Redux & Antd based dashboard
• Yolov3 interpretation script fix and changes to mapping.json
• YOLO format support (#1151)

Fixed

• Exception in Git plugin #826
• Label ids in TFrecord format now start from 1 #866
• Mask problem in COCO JSON style #718
• Datasets (or tasks) can be joined and split to subsets with Datumaro #791
• Output labels for VOC format can be specified with Datumaro #942
• Annotations can be filtered before dumping with Datumaro #994

[0.5.2] - 2019-12-15

Fixed

• Frozen version of scikit-image==0.15 in requirements.txt because next releases don't support Python 3.5

[0.5.1] - 2019-10-17

Added

• Integration with Zenodo.org (DOI)

Added

• A converter to YOLO format
• Installation guide
• Linear interpolation for a single point
• Video frame filter
• Running functional tests for REST API during a build
• Admins are no longer limited to a subset of python commands in the auto annotation application
• Remote data source (list of URLs to create an annotation task)
• Auto annotation using Faster R-CNN with Inception v2 (utils/open_model_zoo)
• Auto annotation using Pixel Link mobilenet v2 - text detection (utils/open_model_zoo)
• Ability to create a custom extractors for unsupported media types
• Added in PDF extractor
• Added in a command line model manager tester
• Ability to dump/load annotations in several formats from UI (CVAT, Pascal VOC, YOLO, MS COCO, png mask, TFRecord)
• Auth for REST API (api/v1/auth/): login, logout, register, ...
• Preview for the new CVAT UI (dashboard only) is available: http://localhost:9080/

Changed

• Outside and keyframe buttons in the side panel for all interpolation shapes (they were only for boxes before)
• Improved error messages on the client side (#511)

Removed

• "Flip images" has been removed. UI now contains rotation features.

Fixed

• Incorrect width of shapes borders in some cases
• Annotation parser for tracks with a start frame less than the first segment frame
• Interpolation on the server near outside frames
• Dump for case when task name has a slash
• Auto annotation fail for multijob tasks
• Installation of CVAT with OpenVINO on the Windows platform
• Background color was always black in utils/mask/converter.py
• Exception in attribute annotation mode when a label are switched to a value without any attributes
• Handling of wrong labelamp json file in auto annotation (https://github.com/opencv/cvat/issues/554)
• No default attributes in dumped annotation (https://github.com/opencv/cvat/issues/601)
• Required field "Frame Filter" on admin page during a task modifying (#666)
• Dump annotation errors for a task with several segments (#610, #500)
• Invalid label parsing during a task creating (#628)
• Button "Open Task" in the annotation view
• Creating a video task with 0 overlap

Security

• Upgraded Django, djangorestframework, and other packages

Fixed

• JavaScript syntax incompatibility with Google Chrome versions less than 72

Added

• OpenVINO auto annotation. It is possible to upload a custom DL model and annotate images automatically.
• Ability to rotate images/frames in the client part (Ctrl+R, Shift+Ctrl+R shortcuts) (#305)
• ReID application for automatic bounding boxes merging (#299)
• Keyboard shortcuts to switch next/previous default shape type (box, polygon etc) [Alt + <, Alt + >] (#316)
• Converter for VOC now supports interpolation tracks
• REST API (/api/v1/*, /api/docs)
• Semi-automatic semantic segmentation with the Deep Extreme Cut work
• Converter to tfrecords format to cvat/utils (#368)
• Support for 16bit grayscale images for annotation (#342)

Changed

• Python Django server design and DB layout
• Propagation input moved from settings to the bottom player panel
• Additional events like "Debug Info" or "Fit Image" for analytics
• LFS is optional for git annotation storages (#314)

Deprecated

• "Flip images" flag in the create task dialog. Rotation functionality in client part added instead.

Added

• Ability to copy Object URL and Frame URL via object context menu and player context menu respectively.
• Ability to change opacity for selected shape with help "Selected Fill Opacity" slider.
• Ability to remove polyshapes points by double click.
• Ability to draw/change polyshapes (except for points) by slip method. Just press ENTER and moving a cursor.
• Ability to switch lock/hide properties via label UI element (in right menu) for all objects with same label.
• Shortcuts for outside/keyframe properties
• Support of Intel OpenVINO for accelerated model inference
• Tensorflow annotation now works without CUDA. It can use CPU only. OpenVINO and CUDA are supported optionally.
• Incremental saving of annotations.
• Tutorial for using polygons (screencast)
• Silk profiler to improve development process
• Admin panel can be used to edit labels and attributes for annotation tasks
• Analytics component to manage a data annotation team, monitor exceptions, collect client and server logs
• Changeable job and task statuses (annotation, validation, completed). A job status can be changed manually, a task status is computed automatically based on job statuses (#153)
• Backlink to a task from its job annotation view (#156)
• Buttons lock/hide for labels. They work for all objects with the same label on a current frame (#116)

Changed

• Polyshape editing method has been improved. You can redraw part of shape instead of points cloning.
• Unified shortcut (Esc) for close any mode instead of different shortcuts (Alt+N, Alt+G, Alt+M etc.).
• Dump file contains information about data source (e.g. video name, archive name, ...)
• Update requests library due to https://nvd.nist.gov/vuln/detail/CVE-2018-18074
• Per task/job permissions to create/access/change/delete tasks and annotations
• Documentation was improved
• Timeout for creating tasks was increased (from 1h to 4h) (#136)
• Drawing has become more convenience. Now it is possible to draw outside an image. Shapes will be automatically truncated after drawing process (#202)

Fixed

• Performance bottleneck has been fixed during you create new objects (draw, copy, merge etc).
• Label UI elements aren't updated after changelabel.
• Attribute annotation mode can use invalid shape position after resize or move shapes.
• Labels order is preserved now (#242)
• Uploading large XML files (#123)
• Django vulnerability (#121)
• Grammatical cleanup of README.md (#107)
• Dashboard loading has been accelerated (#156)
• Text drawing outside of a frame in some cases (#202)