We are pleased to announce the release of the second issue of the InVID newsletter! The aim of this issue is to inform the community, our readers and supporters, about the developed tools and applications for media collection and verification.
Within the pages of this issue, the reader can find information about:
the adopted software development methodology and the progress made so far in the project;
our latest advances over a set of individual technologies for media collection, analysis, retrieval and rights management;
We are happy to announce that a new version of the InVID plugin for video verification was released on November 20th 2017. This version contains a number of new functionalities and improvements that were based on the collected feedback from the users of the tool. In particular, the latest release of the InVID verification plugin:
Integrates an improved version of the video contextual verification tool that supports UTC time conversion, multilingual comment detection and verification and progressive display of results.
Allows the application of reverse image search on YouTube thumbnails through the Bing engine.
Reverse YouTube thumbnail search using the Bing engine (video source: Sputnik)
Includes an extended version of the video keyframe extraction tool that offers additional keyframes for reverse keyframe search.
The video keyframe extraction tool offers additional keyframes for reverse image search (right image) after a user request via a simple button at the end of the initial collection of keyframes (left image)
Provides an improved presentation of the collected image/video metadata.
Integrates Google Analytics for gathering information about its use (related privacy disclaimer has been added in the “About” tab of the tool).
Contains a tutorial that explains the different functionalities of plugin.
During three days, InVID project was demoed by Gerhard Rudinger (APA-IT) and Denis Teyssou (AFP) in the startup space of the World Digital Content and IFRA Expo in Messe Berlin, the traditional yearly event of the press and publishing industry organized by WAN-IFRA, the World association of newspapers and publishers.
The event was the occasion to showcase the InVID project and the InVID Verification Plugin (a.k.a. “Fake video news debunker”) to executives from media houses like News UK (“The Times”, “The Sun”, “Sunday Times”), Kyodo News (Japan), Roularta (Belgium), German newspapers and so on.
InVID was also presented on the main conference stage in the panel “Preparing for the 2020s”, where Denis Teyssou demonstrated the InVID Verification Plugin in front of more than 100 media professionals (see indicative images below):
on a Russian video re-used out of context to create a xenophobic climate during the French presidential election;
on the Eduardo Martins fake war photographer affair, with a new feature aiming to detect pictures which have been mirrored;
on a North Korea picture which shows evidence to have been tampered.
Mr. Denis Teyssou demonstrating the InVID Verification Plugin to the participants of the panel “Preparing for the 2020s” of WAN-IFRA.
Indicative view of the audience of the panel “Preparing for the 2020s” of WAN-IFRA.
The given presentation can be seen in the following:
The main goal of the InVID project “to provide journalists with the tools they need to verify the authenticity and trustworthiness of online videos from news events” was initially highlighted by Dr. Nixon. Following the main integrated technologies of the project, namely the InVID Dashboard, the InVID Verification Plugin and the InVID Verification Application, were presented to an audience comprised by members of other ICT-19 projects, start-ups and investors.
In the video below, you can see the part where the InVID product portfolio was discussed.
The slides of this presentation are also available.
A new release of InVID verification plugin was made on October 3, 2017. This new updated version of the tool contains the following features.
A permanent feedback mechanism that allows our users to get in contact with us in order to share how they use the tool, problems they may have, questions about the applied analysis, as well as ideas for future improvements.
The feedback mechanism of the InVID Verification Plugin
A survey to gather the users’ opinion about the plugin; this survey takes only a few minutes to complete and it will help us to provide you with better tools!
The integrated survey about the use of the InVID Verification Plugin
A new filter that allows users to deal with fake mirrored images. The case of the fake war photographer Eduardo Martins which was reported in early September 2017 puts in evidence that reverse image search engines often fail to detect mirrored images. This use case gave us the idea to propose a tool to overcome such fake images. In the Magnifier feature, there is now a flip filter that allows you to mirror the image before sending it to Google Images. The generated mirrored image can be also downloaded to the user’s PC. See the example bellow and watch the video tutorial on the InVID channel, to see how to proceed.
Magnifier tool for image mirroring
Finally, the same Magnifier feature now support links to images stored in Dropbox and Google Drive (please make sure to have a link ending with “view?usp=sharing”).
Get the latest version of the InVID Verification Plugin from here!
The theme for the 2017 FIBEP World Media Intelligence Congress, to be held in Berlin from 04-06 October 2017, will be Media Intelligence and the New Paradigm of Brand Communication, including a special session on “How to Analyze in an Alternative Facts World”.
The InVID technologies for media discovery and verification will be presented at the ACM Multimedia Conference 2017 that will take place on October 23-27, 2017, at Mountain View, CA USA.
Dr. Lyndon Nixon from MODUL Technology GmbH will present the latest developments of the InVID project on detecting news stories from Twitter streams and using those stories to collect automatically relevant videos from social networks. Specifically, Dr. Nixon will address the need to first find good candidate videos from the haystack of social networks and explain how the InVID Dashboard (see screenshot below) can be used to browse news stories and relevant video posted about those stories, helping journalists more easily identify candídates for verification.
Screenshot of the developed InVID Dashboard
Following, Dr. Vasileios Mezaris from Information Technologies Institute of the Centre for Research and Technology Hellas will focus on the necessity to evaluate the trustworthiness of user-generated videos published on social networks, aiming either to verify their authenticity and veracity or to debunk fake videos. In this context, Dr. Mezaris will demonstrate the use and functionality of the InVID Verification Plugin (see screenshot below) for assessing the reliability of a user-generated video, by investigating the originality, the contextual information, the existence of potential tampering, and the relation to other newsworthy videos.
Screenshot of the Magnifier tool of the developed InVID Verification Plugin
The InVID project and the developed InVID Verification Plugin was presented on Thursday 14th of September at the Brussels DisinfoLab, which has been organized by Saper Vedere, a consulting startup specialized in social networks analysis. At this event, whose sub-headline is «a framework to fight fake news?», various initiatives to debunk fake news were presented from several institutions, labs and research institutes.
Denis Teyssou, the InVID Innovation Manager, demonstrated the use and functionality of the InVID Verification Plugin on recent fake videos published on various social networks. His presentation is available at the InVID SlideShare channel and the project website.
Further details about the context, the aim, the presented items and the next steps of the Brussels DisinfoLab event can be found here.
The European Horizon 2020 project InVID (In Video Veritas) is releasing this Wednesday (12/7) in open beta the Firefox version of its plugin to help journalists save time in their verification processes and debunk more efficiently fake video news. At the same time, the plugin code is published and shared on GitHub (https://github.com/invideu/invid-verification-plugin) in open source with a MIT license. (The plugin code is the same on Chrome and Firefox).
The InVID plugin allows to get contextual information on YouTube and Facebook videos, to fragment videos from several platforms into keyframes, to query reverse image search engines like Google, Yandex and Baidu, to magnify pictures to read details, to read metadata and to apply forensic filters to detect alterations of an image, as well as searching on Twitter by time range up to the minute.
The plugin is available for Windows, Mac OS X and Linux users. To install the plugin, after downloading and unpacking the zip file, simply drag & drop the created XPI file in a new tab of your Firefox browser.