DoxDetect

Our Logo

Inspiration

Whenever we post what we're doing on Instagram, we notice that our friends can easily figure out where we are. While this information is safe with our friends, we fear what would happen if it fell into the wrong hands

What it does

Our app scans through images that we will upload on social media and detects any important information about the user. It then prompts the user to choose whether they want to blur that information or not. This way, our users can look great without the worry of losing their privacy

How we built it

We built our program through python EasyOCR detected text within images OpenCV highlighted that text Regex and OpenAI determined if that text was "personal information" or not. If so, the users were given an option to blur it. Tkinter was used to create the GUI to demonstrate our program

Challenges we ran into

We originally used Pytesseract instead of easyOCR, but once we discovered that easyOCR didn't detect numbers, we switched over to Pytesseract. Researching and discussing which resources to use and how to implement those resources was a challenge that we constantly fought through during the entirety of our app's creating

Accomplishments that we're proud of

We accomplished developing our very own app that can be deemed useful even in our own lives. The app we created took a lot of work, and we are proud of what became of it

What we learned

We learned how to use easyocr, and a lot about optical character recognition. We also learned some about tkinter, which taught us basic skills on how to create a GUI.

What's next for DoxDetect

Improved UI - The UI we had today was more of a proof of concept than an aesthetic user interface Training our own AI models - OpenAI did a great job, but there was some mistakes. By training our AI models, our program can become more effective More languages - We can only do English right now, but most of the world are non-english speakers. By introducing detection of more languages, we can scale our app furthur