Sus or Safe

Discord - makshiboi#0210

Inspiration

We knew that we wanted to take advantage of some Google APIs, so we brainstormed some projects that we could do with them. We thought of this idea as a way to combat websites that do not already have integrated content moderation in order for web surfers to 'explore the unknown.'

What it does

Sus or Safe scans a website's images and text to let the user know whether its content is sensitive. If the website has potentially explicit/offensive content, the user will be alerted. If the content is safe, the user will also be informed. Either way, stats relating to the website's content will also be displayed.

How we built it

We made a variant of a web-scraper in Python to collect images from the HTML of a given URL. Those images were sent to a bucket in Google Cloud Storage. The images were then processed with Cloud Vision to determine the content of the images. Additionally, the text of the website gets collected and ran through Natural Language Processing to determine what the text is about.

Challenges we ran into

It was hard to figure out how to implement the Google APIs, especially given that we chose to use Python, a language we were unfamiliar with. It was also difficult to gather images from just the HTML.

Accomplishments that we're proud of

Being able to develop familiarity with Python while simultaneously implementing sophisticated, enterprise-level, AI API libraries.

What we learned

We learned how to use Google Cloud Vision, Cloud Storage, and Natural Language Processing.

What's next for Sus or Safe

Our image retrieval algorithm can be optimized to filter out some images as well as discover more images not found through HTML currently. Sus or Safe doesn't handle websites that require login information or other user data very well, but there might be a way to fix that. We can also improve our usage of Google Cloud Storage and upgrade our app UI into something more dynamic. We hope to continue growing our project with these initiatives in mind.

Built With

Submitted to

Swamphacks IX

Created by

I developed the functionality of moving the images stored locally into a cloud bucket using the Cloud Storage API. I also made the system for which images are scanned using Google Cloud Vision, and I organized the data that would be produced from all images found in a website.

Max Huang
I integrated the Google Cloud Natural Language API in order to implement NLP functionality so that Sus or Safe could read the text of a webpage and determine its topic. I also built a simple web parser using BeautifulSoup to grab the text from webpages.

Tony Gupta
I worked on web scraping images from various websites and saving those images into a temporary local folder.

Liam Sy
Brenda Nakasone