Multimodal Fusion based Adult Website Detection

The internet is a vast source of information, but it also contains a significant amount of adult content that is not appropriate for children, schools and office environments. The problem of such content on the internet is addressed by utilizing deep learning models to identify and block Not Safe For Work(NSFW) websites. Current firewalls lack in detecting websites with only images, text or misleading domain names. This proposed solution uses a multimodal weighted late fusion technique, combining features of the website’s source code and images, to classify a website as safe or NSFW. Firewalls can use the proposed solution to effectively block NSFW content to provide internet security and maintain a healthy and productive environment.