junctionTokyo19
Inspiration
The proliferation of obscene contents, as well as online harassments in the cyber world, is becoming concerning day by day with the rise of social media. Though pornographic websites can be blocked easily, blocking disturbing contents on social networks and seemingly benign websites is becoming challenging from the end user perspective. Our motivation for this project is the empowerment of the users to control the contents they view.
What it does
The purpose of this project is to build a google chrome extension which will use Artificial Intelligence to detect obscene contents (images and texts for now) in any website and blur out that portion.
How we built it
Firstly, we have trained two deep learning models, one for images and another for texts. These models are capable of distinguishing obscene content from normal contents. Secondly, we have built a google chrome extension with Java-script which would scan all the content of the website before showing up. Finally, we incorporated the deep learning models with the extension so that, the extension will get to know whether the content is obscene or not from the deep learning models.
Challenges we ran into
A browser extension is a client-side application, we had limitations to access data for security issues. Another challenge was to loading the model as we made the prediction of contents on the client end with Java-script.
Log in or sign up for Devpost to join the conversation.