/ME2016_unitn

Primary LanguagePythonOtherNOASSERTION

ME2016_unitn

In this project, we developed a hybrid approach for verifying multimedia use on Twitter. This work converged into the manuscrit available at this link and presented at MediaEval 2016.

The project contains below main parts:

  1. Image_Forensics: extract forensic features from images in development set and test set.

  2. Text_Retrieval: perform google image reverse search and extract relevant websites containing images similar to the query image. To videos, extract users' comments from Youtube.

  3. Textual_Feature: extract useful textual features from text retrieved in part 2.

  4. Tweet features: from a user's post, extract post-based and user-based features.