CENSORSHIP TOOL TO DETECT NSFW CONTENT IN A VIDEO FILE

Abstract

Internet has vast content which is spread like wildfire across different devices, countries, cultures and ages. When so much content is available it is often prone to abuse. The abuse is mainly from NSFW (Not Safe For Work) which reaches inappropriate audiences.  Our project focuses on detecting the NSFW content and letting user know if the content that they are viewing has NSFW before viewing the content. Our idea is to work with video files as they possess NSFW frames which popup without user’s control over it. We use a pre-trained classifier (nudenet) to classify if a frame has NSFW content or not.