Detection, characterization and analysis of unsafe video content on YouTube

A safe video platform can provide a healthy and suitable environment for users in general and children in particular. This project aims to utilize machine learning and deep learning techniques to identify and flag sensitive and questionable content (e.g., content related to violence, sexuality, etc.). The algorithm will leverage video frames extracted from the database for training and building a 3D Convolutional Neural Network (CNN) model. This model can detect and classify videos just like a human being. With this technology, BBTV can provide a more appropriate channel management solution to content creators that leverage its services.

Faculty Supervisor:

Fengjun Yan

Student:

Yangliu Dou

Partner:

BroadbandTV Corp.

Discipline:

Engineering - mechanical

Sector:

Information and communications technologies

University:

Program: