Abhishek Dutta, Ankush Gupta and Andrew Zisserman



Overview

VGG Image Annotator is a simple and standalone manual annotation software for image, audio and video. VIA runs in a web browser and does not require any installation or setup. The complete VIA software fits in a single self-contained HTML page of size less than 400 Kilobyte that runs as an offline application in most modern web browsers.

VIA is an open source project based solely on HTML, Javascript and CSS (no dependency on external libraries). VIA is developed at the Visual Geometry Group (VGG) and released under the BSD-2 clause license which allows it to be useful for both academic projects and commercial applications.

Demonstrations

Click on the images to try the VGG Image Annotator application preloaded with images and annotations.

VIA Demo Screenshot
Basic image Annotation
VIA Video Annotation
Video Annotation
VIA Audio Annotation
Audio Annotation
VIA Face Annotation Demo Screenshot
Face Annotation
VIA Face Track Annotation
Face Track Annotation
Video Subtitle Annotation
Video Subtitle Annotation
Video Subtitle Annotation
List Annotator (LISA)

Downloads

Version 2
image annotator


Version 3
audio and video annotator


VIA version 1 is an early version of VIA software and many users still find it useful for basic image annotation tasks.
List Annotator (LISA) : to efficiently annotate a large list of images.

Note: We have tested this application in the following browsers: Firefox, Chrome and Safari. Our users have reported that this application also works well in other browsers like Internet Explorer, GNOME Web, etc. .

User Guide

A basic user guide is also available inside the VIA application and can be accessed using top menubar "Help -> Getting Started". Most common user queries about the VIA software are available in our Frequently Asked Questions (FAQ) page.

Citation

If you use this software, please cite it as follows:

Abhishek Dutta and Andrew Zisserman. 2019. The VIA Annotation Software for Images, Audio and Video. In Proceedings of the 27th ACM International Conference on Multimedia (MM ’19), October 21–25, 2019, Nice, France. ACM, New York, NY, USA, 4 pages. https://doi.org/10.1145/3343031.3350535.

@inproceedings{dutta2019vgg,
  author = {Dutta, Abhishek and Zisserman, Andrew},
  title = {The {VIA} Annotation Software for Images, Audio and Video},
  booktitle = {Proceedings of the 27th ACM International Conference on Multimedia},
  series = {MM '19},
  year = {2019},
  isbn = {978-1-4503-6889-6/19/10},
  location = {Nice, France},
  numpages = {4},
  url = {https://doi.org/10.1145/3343031.3350535},
  doi = {10.1145/3343031.3350535},
  publisher = {ACM},
  address = {New York, NY, USA},
}

@misc{dutta2016via,
  author = "Dutta, A. and Gupta, A. and Zissermann, A.",
  title = "{VGG} Image Annotator ({VIA})",
  year = "2016",
  howpublished = "http://www.robots.ox.ac.uk/~vgg/software/via/",
  note = "Version: X.Y.Z, Accessed: INSERT_DATE_HERE"
}

Developer's Resources

VIA software is developed using HTML, CSS and Javascript and is based solely on standard features available in modern web browsers. VIA does not depend on any external libraries. These design decisions has helped us create a very light weight and feature rich manaul annotation software that can run on most modern web browsers without requiring any installation or setup. The full VIA software sprouted from an early prototype of VIA which implemented a minimal -- yet functional -- image annotation tool using only 40 lines of HTML/CSS/Javascript code that runs as an offline application in most modern web browsers. This early prototype provides a springboard for understanding the current codebase of VIA which is just an extension of the early prototype. Furthermore, the introductory tutorials prepared by Mozilla is also very helpful in understanding the basic concepts of HTML/CSS/Javascript platform.

The VIA source code repository contains a separate folder for each major version of VIA: via-1.x.y, via-2.x.y and via-3.x.y. The development of each version is carried out in a separate branch (e.g. via-2.x.y branch). If you wish to contribute code to VIA (we encourage you to do so), please send a pull request to one of the branches. Please do not send pull requests to the master branch. Furthermore, all contributions must adhere to the existing license of the VIA project. The following code documentation are available for each major version of VIA:

Scripts

The following scripts are often useful for processing the annotations produced by the VIA software. If you have such scripts that you would like to share with other users, please let us know.

Version Script Description Dependencies
via-3.x.y CDLI COCO to Web Annotation Format (External) First convert VIA3 annotations to COCO format and then to Web Annotation Data Model format using the coco_to_w3c.py script developed by the CDLI project. Python3
via-2.x.y extract_img_regions.py Crop and save image regions defined in a VIA-2.x.y project (json file) Python3, PIL Image
via-2.x.y merge_via2_projects.py Merge multiple projects (json file) created using VIA-2.x.y Python3
via-3.x.y merge_via3_projects.py Merge multiple projects (json file) created using VIA-3.x.y Python3
via-3.x.y export_via_annotations.py Export VIA3 shared project annotations in CSV format Python3
via-3.x.y via3_video_annotation_to_coco.html Export VIA3 shared project frame level video annotations in COCO format View HTML in Firefox browser

Contact

Contact Abhishek Dutta for any queries or feedback related to this application. We welcome all forms of contributions (code update, documentation, bug reports, etc) from users. These contributions must adhere to the existing license of VIA project.

News

See CHANGELOG for more details.

  • VIA 1.0.7 : (02-Jul-2024) maintenance release of VIA 1.0.x series with a minor fix.
  • VIA 3.0.13 : (02-Jul-2024) bug fixes and improvements for video frame region annotations, see CHANGELOG for more details.
  • VIA 3.0.12 : (14-Jul-2023) bug fixes with temporal segmenter in video annotator.
  • VIA 2.0.12 : (26 Sep. 2022) point region now gets assigned default value, other minor fixes.
  • FAQ: (09-Feb-2022) added Frequently Asked Questions page.
  • LISA 0.0.3 : (08-Oct-2021) public release of List Annotator (LISA) which can be used to efficiently annotate a large list of images or videos.
  • VIA 3.0.11 : (08-Oct-2021) temporal segments can now have attributes, add/delete polygon vertex, added magnifier and zoom for image annotator.
  • VIA 3.0.10 : (04-March-2021) lots of bug fixes and some minor improvements, see CHANGELOG.
  • VIA 2.0.11 : (03 Mar. 2021) Polygon vertices can now be added/deleted using [Meta] + Click operation in MacOS, status messages can now be disabled.
  • 10 Nov. 2020 : Slides (pdf, odp) and videos (mp4, youtube) from a talk titled "Learnings from developing and maintaining a research software that has been used more than 3 million times in the last 3 years" at International Series of Online Research Software Events (SORSE) on Nov. 10, 2020.
  • 03 Nov. 2020 : Slides (pdf, odp) and videos (mp4, youtube) from presentation titled "Manual Annotation of Images and Video using VIA" at VGG Meeting on Nov. 3, 2020.
  • VIA 3.0.9 : (09 Oct. 2020) added subtitle editor for audio and video, added zoom (i.e. magnifying glass) feature to inspect finer details in image annotator.
  • VIA 3.0.8 : (01 July 2020) added standalone external HTML application to export frame level video annotations to COCO format, temporal segment rows are sorted numerically, audio amplitude visualization for video annotation (experimental feature, disabled by default), fixed other issues.
  • VIA 2.0.10 : (08 Jun. 2020) COCO import/export code were rewritten to address multiple issues reported by users, other user reported issues were also fixed.
  • VIA 3.0.7 : (07 May 2020) improved timeline in video and audio annotation, fixed some issues.
  • VIA3 JSON Format : (01 May 2020) Description of the structure of VIA3 Project JSON File.
  • Pair Annotation : (16 Jan. 2020) VIA Use Case: Pair Annotation shows how VIA3 can be used to define metadata associated with an image pair.
  • VIA 2.0.9 : (02 Mar. 2020) bug fixes release, see CHANGELOG.
  • ACM MM 2019 : (24 Oct. 2019) we presented a paper about the VIA software at the ACM Multimedia Conference 2019.
  • VIA 3.0.6 : (20 August 2019) collapse file and spatial metadata editor, spatial regions are automatically associated with selected temporal segments. (updated previous 3.0.6 release to included minor fixes)
  • VIA 3.0.5 : (14 August 2019) extreme clicking method for rectangular and circular regions, import VIA2 projects, file metadata editor, three way merge for VIA shared projects, etc.
  • VIA 3.0.4 : (02 July 2019) region label, bulk file URI addition, fixed major issue with file-id and view-id, minor improvements and bug fixes.
  • VIA 3.0.3 : (26 June 2019) collaborative annotation, resolve file properties, bug fixes.
  • VIA 2.0.8 : (14 June 2019) rotated ellipse, import/export of COCO annotations, shows region shape description and bug fixes.
  • arXiv:abs/1904.10699 : (31 May 2019) VIA now supports manual annotation of Images, Audio and Video. We updated our arxiv paper to describe recent updates to the VIA suite of open source applications.
  • VIA 3.0.2 : (21 May 2019) bug fixes (shape selector icon gets highlighted, fixed temporal segment variable value rename, minor UI updates)
  • VIA 3.0.1 : (20 May 2019) fixed issue with keyboard shortcuts.
  • VIA 3.0.0 : (17 May 2019) A major upgrade to the VIA codebase to allow manual annotation of audio and video.
  • VIA 2.0.7 : (01 May 2019) user's mouse cursor coordinate is visible (172), some bug fixes (issue 173, merge 15).
  • arXiv:abs/1904.10699 : (25 Apr. 2019) A report describing the VGG Image Annotator (VIA) is now available from arXiv.
  • VIA 2.0.6 : (25 Mar. 2019) regions can now be coloured based on attribute value, some bug fixes.
  • MPLS Award 2019 : (19 Feb. 2019) Dr. Abhishek Dutta was awarded the MPLS Early Career Impact Award for developing "the VGG Image Annotator – a widely used open source manual image annotation software application".
  • VIA 2.0.5 : (03 Dec. 2018) fixed issue which prevented firefox users from downloading project, annotations, etc.
  • VIA 2.0.4 : (29 Nov. 2018) keyboard shortcut update, bug fixes, selective exclusion of groups images in grid view.
  • VIA 2.0.3 : (25 Oct. 2018) improved keyboard shortcut keys, bug fixes (on-image annotation editor and checkbox attributes).
  • A blog post describing VIA and its open source ecosystem published at VGG blog on 17 Oct. 2018.
  • VIA 2.0.2 : (10 Oct. 2018) quick update of annotations using on-image annotation editor (based on user contribution), bug fixes.
  • VIA 2.0.1 : (26 Sep. 2018) bug fixes and minor improvements (download as image, highlight row in annotation editor, etc.)
  • VIA 2.0.0 : (15 Jun. 2018) major upgrade to VIA with following advanced features: upgraded user interface, projects, image grid view for bulk update of annotations, new attribute types (checkbox, radio, image, dropdown), annotate remote images, new attribute editor and upgraded annotation editor.
  • VIA 1.0.6 : (15 Jun. 2018) maintainenace release of VIA 1.0.x series with minor fixes.
  • VIA 1.0.5 : (16 Jan. 2018) fixed issue 85, merged contributed code (see Contributors.md)
  • VIA 1.0.4 : (17 Oct. 2017) polygon copy/paste issue fixed (see issue 107)
  • VIA 1.0.3 : (07 Aug. 2017) fixed a major issue (extra comma in csv export, see issue 103)
  • VIA 1.0.2 : (04 Aug. 2017) minor bug fix release on August 04, 2017
  • VIA 1.0.1 : (11 Jul. 2017) minor bug fix release
  • VIA 1.0.0 : (04 Apr. 2017) first public release
  • VIA 0.1-beta : (Sep. 2016) early prototype of VIA software used for internal tests

Acknowledgements

Development and maintenance of VIA software has been supported by the following two research grants:

  • Visual AI: An Open World Interpretable Visual Transformer (UKRI Grant EP/T028572/1)
  • Seebibyte: Visual Search for the Era of Big Data (EPSRC Grant EP/M013774/1)

EPSRC logo Seebibyte Logo