Changelog

February 2022 Update

February 28, 2022
A series of cards from V7 Darwin

Introducing: The New Model Tool

Auto-Annotate is getting a major power-up: Use any model to automatically label polygons, bounding boxes, text, and tags. The Model Tool lets you map classes to models from the internet, and will surface confidence scores for every label created.

Confidence scores within the Annotation interface

See confidence scores for every annotation, spot ambiguous objects and set confidence scoring thresholds to control the accuracy of your automatically-labelled data. 

Support for all model types 

Whether it's one of your team's trained object detection or classification models, or one of our public models like the text scanner, the Model Tool can output any annotation type. You may even use it at lower complexity levels, for example use an instance segmentation model to produce bounding boxes.

Label specific regions or entire files

Create multiple unique labels at once by drawing a bounding box over several objects, or run a model over an entire file by hitting Cmd + Enter.

Model Tool also works on ultra-large files, medical imaging, and PDFs.

Triage hard tasks with AI Models 

One of our favorite ways of using Model Tool is to pre-process data and flag low-confidence boxes. This is a great way to flag out-of-distribution objects and direct reviewers to them. You may also use multiple model versions to compare their score on the same image.

More Features and Improvements

Feature
Feature
Feature
Feature
Performance
Performance
Performance
Performance
Fix
Fix
Fix
Fix
Fix
Fix
Fix

New Features

🆕 Label segments can now be easily moved/resized in video timelines when annotating on touch displays.

Performance Improvements

🚀 We've made zooming and panning in multiple-gigabyte TIFF files significantly smoother.

🚀 DICOM series and their slices now have significantly-improved loading times and playback speeds.

Fixes

👾 New annotations now always appear at the top of the Annotations menu.

👾 Corrected the alignment of rendered frames in streamed videos with the frame in the video timeline.

👾 Fixed an issue that prevented Auto-Annotate from functioning in videos with streaming enabled.

👾 Corrected the interpolation algorithm for Ellipse annotations to ensure smooth interpolation between key frames.

👾 Fixed an issue that caused the layering of annotations in the Annotations menu to be inverted when compared to the layers in the exported JSON file.

👾 Tweaked interpolation behaviour to prevent complex polygons (polygons with multiple parts) from disappearing in interpolated frames.

👾 Files in a Complete state no longer display any annotation tools to reflect the fact that they are uneditable.

👾 Fixed an error that prevented DICOMs with hanging protocol layouts from being edited.

👾 Class filters now reset when moving between datasets.

👾 All class names were previously capitalised in the workview. This is no longer the case.

👾 Keypoints now change appearance when hovering over them with the cursor. This makes them clearer when not the cursor is not hovering above them.

January 2022 Update

January 28, 2022
A series of cards from V7 Darwin

Annotate videos with audio (beta)

 

Do you hear that? That's the sweet sound of a new beta feature. Team, Business, and Enterprise tier accounts can now test our updated video platform in beta which includes the ability to annotate with audio. Team owners, shoot us a message at support@v7labs.com to enable video streaming for your team. All files imported after the update will play back at full speed with audio. 

Image Settings Presets

 

Save DICOM windowing levels and other settings from the Image Manipulation Panel with Image Settings Presets. Quickly apply your saved settings using hotkeys and avoid having to tinker with windowing levels ever again. 

FDA Part 11 Compliance  

 

V7 is now compliant with the United States Food and Drug Administration’s (FDA’s) 21 CFR Part 11 regulation.

Now that V7 can prove that its annotations comply with Part 11, the FDA will accept their electronic signatures, allowing doctors to collaborate on our platform in an entirely paperless and cloud-based fashion.

Smoother video annotation

 

A series of updates has optimized video rendering, sped up frame load-times, and made playback and jumping to unloaded frames faster than ever.

More Features and Improvements

Feature
Feature
Feature
Feature
Performance
Performance
Performance
Performance
Fix
Fix
Fix
Fix
Fix
Fix
Fix
Performance Improvements
🚀 Improved performance of color maps in the Image Manipulation panel.
🚀 Smoother panning when annotating while zoomed in.

Bug Fixes

🛠 Added a fix to ensure backend stability if unexpected data types are used as dataset IDs.
🛠 Fixed the filter parameters for workflow stage filters in the Data page.
🛠 Bulk archives no longer archive all items when an invalid filter is selected.
🛠 Fixed an error that prevented unused credits from rolling over upon upgrade in some cases.
🛠 Keypoints now respond to annotation opacity settings in the Image Manipulation panel.
🛠 Added an error message when marking files as Complete in bulk using the SDK fails.
🛠 Fixed an issue that caused exports in the Coco and Semantic Mask PNG formats to get stuck in Creating version.
🛠 Fixed an issue that caused the settings in the Image Manipulation panel to reset with each new file.
🛠 Fixed a display issue that caused text to overlap in the Plans page of the Settings menu.
🛠 Fixed an issue preventing cuboids from rendering properly in video files.
🛠 Copying annotations between two videos now copies video key frames.
🛠 Fixed an issue that prevented holding the centre mouse button from being used for panning.
🛠 Fixed an issue that caused the comment box to appear on consecutive frames.
🛠 Fixed an issue that caused DICOM files to enter an "Error" state during processing.
🛠 Fixed an issue preventing DICOM registration from external storage to fail for some files.
🛠 Fixed an issue preventing teams from accessing reports via the API.
🛠 Added negative width/height support for BMP files to prevent some files from entering an "error" state upon import.
🛠 Fixed an issue that prevented annotations from rendering when comments were present.
🛠 Fixed a display issue that caused overlap in the Classes menu.
🛠 Fixed an issue that prevented Auto Annotate from working in tiled files.
🛠 Fixed an issue that prevented subsequent frames from loading when the playhead is dropped in a sequence of unloaded frames.
🛠 Fixed an issue that prevented filter parameters from applying to the file carousel in the workview.
🛠 The keyboard shortcut for editing a polygon with Auto Annotate (Shift + N) now works (it was previously unresponsive).

December 2021 Update

December 22, 2021
A series of cards from V7 Darwin

(Beta) Annotate with your own model

 

In addition to being able to label with models that are trained in V7, customers on our Enterprise plan can now register models hosted on their own architecture, and use them in labelling workflows.

Train and run models using annotation credits

 

Where we used to charge separately for training, hosting, and labelling with V7 Neural Networks, all of V7's Neural Networks features now use your team's balance of annotation credits.

Import annotations from Labelbox

 

We've added Labelbox's annotations to the list of formats supported by our SDK to make the transition to V7 even smoother. Labelbox annotations can be imported via the CLI or Python Library. 

More Features and Improvements

Feature
Feature
Feature
Feature
Performance
Performance
Performance
Performance
Fix
Fix
Fix
Fix
Fix
Fix
Fix

New Features

🆕Users and use roles can now be assigned via SSO.

🆕Added support for the NDPI format.

🆕The Instructions panel in the Settings page of any dataset now supports image, video, and .pdf files.

Fixes

🛠 It's no longer possible to create annotation segments with a negative value in JSON exports.

🛠 Fixed an issue causing duplicate labels in some cases. 

🛠 Fixed an issue in the frontend that caused annotations with a length of 0 frames.

🛠 Added support for the same tag class to be imported at separate points in a video's timeline. 

🛠 Fixed an issue that caused inconsistency in the layering in the Annotations menu.

🛠Applied a few fixes to the auto-start feature when starting a Neural Network to increase consistency.

🛠Applied a fix to one of our backend workflow processes that caused performance issues when a stage that was supposed to complete automatically was stuck. 

🛠Fixed a related issue that caused automatically-created annotations in a workflow stage to crash. 

🛠Fixed a frontend issue that prevented single annotations in the Annotations menu from expanding so that attributes could be added.

🛠The precision/recall matrix for trained Neural Networks now only appears on applicable models.

🛠Fixed a display issue that prevented some annotations from appearing when "travelling back in time" (moving back one workflow stage).

🛠Removed overlap between the video timeline and image canvas for video and DICOM files. 

🛠When creating a hanging protocol layout, the miniature layouts that appear in the Annotations menu now match the layout in the image canvas.

Performance Improvements

🚀 Added a new way of indexing our database which will unlock several performance improvements.

🚀 Annotations now only render in the field of view that's visible to the user, saving loads of memory.

🚀 The Annotations menu now loads much faster, particularly in files that are dense with annotations.

November 2021 Update

November 24, 2021
A series of cards from V7 Darwin

Create hanging protocols with DICOM layouts

 

Our latest beta feature, DICOM Layouts, gives medical teams the ability to create custom DICOM hanging protocols in three different layout types: Grid, Horizontal or Vertical.

Annotate with three new color maps


The Image Manipulation panel now includes three additional color maps to make every detail in your data perceptible: Hot, Seismic, and Paired.

Configure external storage from the UI 

Admins can now configure AWS S3, GCP, Azure Blob Storage, and MinIO integrations from Settings > Storage.

Reveal class performance through Confusion Matrix

 

Trained classification models now include a Confusion Matrix to help you better understand your model's performance.‍

More Features and Improvements

Feature
Feature
Feature
Feature
Performance
Performance
Performance
Performance
Fix
Fix
Fix
Fix
Fix
Fix
Fix

New Features

🆕 Added support for .rvg file imports.

🆕 When deleting a class across several datasets, a notice now appears with additional controls to ensure that annotations are not deleted in error.

🆕 Archived files are now overwritten when uploading new files that contain the same filename.

 

Fixes

🛠 Fixed an issue that prevented some labels from appearing on the image canvas.

🛠 Fixed an issue that intermittently prevented comments from appearing before a refresh of the page.

🛠 When sipping together several DICOM files, imports now support uppercase .DCM file extensions in the .zip folder.

🛠 Batch-completing files was not updating the updated_at endpoint when using the REST API to get file details.

🛠 Fixed an issue that caused annotations to be removed from the image canvas when only one image contained a tag was sent through a consensus workflow.

🛠 Fixed an issue that caused some annotations in video files to have a length of 0 frames. 

🛠 Fixed two issues which caused crashes when using specific filters in the REST API.

🛠 Fixed a bug where temporary data wasn't cleaned, causing some of our servers to run out of memory.

🛠 Fixed an issue preventing batches of files with large amounts of annotations from being bulk completed.

🛠 Fixed two issues impacting partner teams, where members of parent teams and child teams could not be added/removed from workflow stages.

🛠 Introduced a fix to prevent the import and creation of annotations on video files that contained an invalid segment length.

🛠 Fixed an issue that prevented members of a child team from seeing the datasets that they were assigned to until they refreshed the page. 

🛠 Fixed an issue that prevented tags from being applied in bulk when workflow filters were applied. 

🛠 Fixed a display issue that caused some elements to flash when assigning files.

🛠 Added a new, more descriptive error message for when a node or edge of a Keypoint Skeleton class is missing when hitting Save.

🛠 Added an empty state for the Classes page of a dataset when no classes have been added on a team level.

🛠 Fixed a small display issue which caused some elements which should have been static in the main side menu to be scrollable.  

 

Performance Improvements

🚀 We've added faster scaling up of our workers to improve video processing speed.

🚀 Folders and item counts now render faster in the Data page of a dataset.

🚀 Searching for classes in teams with over 1000 classes is much faster.

🚀 Curser movement in the workview has been made smoother.

🚀 Exports containing several million annotations now process in less than 20 minutes.

October 2021 Update

October 26, 2021
A series of cards from V7 Darwin

Train an Object Detection Model

Train an object detection model from within V7 on any dataset with at least 100 instances of the polygon or bounding box classes that you wish to train on. Easily set up a model-assisted workflow and automatically create thousands of bounding box labels in minutes. 

Manage Multiple Teams with Partner Teams

 

V7's Enterprise plan now gives you the ability to manage multiple, separate "child" teams from a single "parent" team.

Auto-Start Your Models

 

Model-assisted workflows have never been easier. Enable Start when invoked to automatically spin up a trained model as soon as a request is made. 

Integrate with Azure and GCP Storage

 

Slotting in alongside AWS S3 storage, V7 has added the option to store and register files from Azure and GCP buckets for teams on our Business plan and above.

Sign in With Google

It's never been easier for team members with Google accounts to securely log into V7. 

More Features and Improvements

Feature
Feature
Feature
Feature
Performance
Performance
Performance
Performance
Fix
Fix
Fix
Fix
Fix
Fix
Fix

New Features

🆕 Downloads of the .csv report in the Overview page of any dataset are now version-controlled. 

🆕 Webhook payloads for completed files are now harmonized with the standard Darwin JSON format. 

🆕 DICOM and video files are now supported for use in workflows with consensus stages. 

🆕 Fetching dataset information via the CLI now reveals a progress bar. 

 

Fixes 

🛠 Fixed an issue that temporarily prevented some of the Support/Feedback button from being clickable.
🛠 Fixed a display issue that caused incorrect user roles to appear in the Team page under Settings. 
🛠 Fixed an issue that temporarily caused Workforce Managers to no longer have access to the Data page of assigned datasets.
🛠 Fixed an issue in the Annotators page that caused the download button to disappear.
🛠 Pre-emptively wrote a fix to ensure that video subannotations are always included in the correct section in Darwin JSON exports.
🛠 Fixed an issue that prevented DICOM files that were not zipped together into a study from importing succesfully.
🛠 Fixed a recurring 500 error when processing the thumbnails of files imported from Azure Blob Storage.
🛠 Fixed an issue that prevented subannotations (Instance IDs and Attributes) from uploading alongside imported annotations.
🛠 Fixed an issue that prevented some invitations to create new trials with V7 from being successfully accepted.
🛠 Removed the previous statuses of archived files from the Dataset Overview Report, making the report easier to interpret.
🛠 Fixed an issue that caused duplicate classes to be created during an annotations import. Followed up the fix with a process that merged any duplicate classes.
🛠 Debugged our internal Model status channel to improve monitoring of active and training models.
🛠 Made a change to display the first frame of a video as the file’s thumbnail.

🛠 Fixed an issue that prevented invitations to join a V7 team from being accepted if already signed into another team. 

🛠 Fixed an issue that only allowed single characters to be typed into the name field when editing a workflow stage. 

🛠 Fixed an issue that caused tags to be removed from files between workflow stages. Identified all affected teams, notified them, and restored the missing files.

🛠 Fixed an issue that temporarily caused a small handful of files related to the issue above to become uneditable. 

🛠 Fixed an issue that caused videos that were supposed to be uploaded as individual images to be opened as regular playback videos. 

🛠 Fixed an issue that caused some annotation exports to be stuck in a Creating version status.

🛠 Fixed an issue that caused some tags to be removed when a file was sent through a consensus workflow stage.

 

Performance Improvements

🚀 Numerous cleanups and changes to our backend processes for faster development and smoother performance.