Warning: opendir(/var/www-wpmu/htdocs/wp-content/cache/db/34/options//bcc/04d): Failed to open directory: No such file or directory in /var/www-wpmu/htdocs/wp-content/plugins/w3-total-cache/Util_File.php on line 158
Multimedia Data Processing - MMSP

Multimedia Data Processing

MULTIMEDIA DATA PROCESSING

Multimedia content has become pervasive. In this research area we mainly focus on the detection of hateful content, especially misogynistic, in viral contents such as memes. Moreover, we work towards the understanding of novel learning approaches by analyzing video lectures and provide a better insight on the user perceived interest in visual contents.

Meme analysis

To learn more contact

Francesca Gasparini

Multimedia data processing - MAMI Research Sub-Topic Supervisor

Multimedia content has become pervasive. In this research area we mainly focus on the detection of hateful content, especially misogynistic, in viral contents such as memes. Moreover, we work towards the understanding of novel learning approaches by analyzing video lectures and provide a better insight on the user perceived interest in visual contents.

Data collection

Collect up-to-date online content.

Misogyny perception

Understand how misogyny is perceived by a heterogeneous population.

Define models

Define learning models considering a multimodal approach for multimedia content analysis, while exploiting natural language processing and image processing algorithms.

Video interestingness

To learn more contact

Claudia Rabaioli

Multimedia data processing - Interestingness Research Sub-Topic Supervisor

Video interestingness is often studied to boost user satisfaction. The goal is to create effective models that predict content engagement using deep learning, but these models are usually difficult for humans to interpret.

Particularly ...

Features

We consider the use of handcrafted features to analyze signal behavior.

Saliency

We use video saliency techniques to identify the most important frames in the videos.

Multimodality

We incorporate both audio and visual analysis, rather than focusing solely on visual data, to enhance our approach