Multimedia Knowledge and Social Media Analytics Laboratory

About InVID

CERTH in InVID - Video Verification

A 5-minute Guide to the Project Outcomes by the MKLab Team

Logo Detection

Image and video logo detector

In UGC video verification, one potential important piece of information is the video origin. This service is able to identify logos in videos, drawing from a large number of sources of TV channels, independent media organizations, and informal groups such as militant organizations participating in the Syrian civil war. The service aims to assist investigators by automatically detecting the logo and providing the users with information about it, such as the name of the organization and the corresponding Wikipedia page. Thus, the investigator may be aware of the video provenance, and be aware of potential biases or manipulation of the truth.

Demo
Contextual Verification

Context Aggregation and Analysis

This is a demo platform aimed to facilitate the verification of UGC video content posted on YouTube, Twitter and Facebook. In contrast to other approaches, which attempt to analyze the videos themselves for traces of forgery, this platform analyzes the video context: The characteristics of the poster, any relevant user comments, the local weather reports at the time of the event, and other contextual pieces of information are aggregated and presented to the user for analysis.

Demo

Publication: Web Video Verification using Contextual Cues

Dataset: Fake Video Corpus

Near Duplicate Detection

Near-Duplicate Video Retrieval

The primary goal of this service is to assist the verification of UGC video content by detecting the similar segments between videos. A significant amount of videos posted in mainstream social media platforms has been analysed and added to a massive video index. The service provides all the important functionalities to the users in order to manipulate and query this video index and retrieve similar videos. In particular, providing the service with a given query, it retrieves all the relevant videos found in the video index and serves them ranked based on their similarity to the query.

Publication: Near-Duplicate Video Retrieval with Deep Metric Learning

Video Verification