Inspiration ☁️:
Our initial inspiration came from the infamous desktop goose “virus”. The premise of the app revolves around a virtual goose causing mayhem on the user’s computer screen, annoying them as much as possible. We thought it was really interesting how the goose was able to take control of the user’s peripheral inputs, and decided to base our project around a similar concept, but to be used for goodwill. Of course, the idea of creating a goose app was certainly enthralling as well. Unfortunately, our designer mistook a goose for a duck, so some plans had to be changed (probably for the better). Currently, one of the most common cybercrimes is phishing (social engineering attack to steal user data). Certain groups, especially children, are often vulnerable to these attacks, so we decided to create a duck companion, one that will protect the user from malicious phishing content.
What it does💁♂️:
"Detective Duck" is a real-time phishing detection application that prevents children from clicking misleading links on the web. It uses a combination of machine learning, cloud technologies, and various python libraries to determine fraudulent links, and where they are located on a page.
Challenges we ran into 😳:
- Examining and weighing bias during the preprocessing step of the data
- Web scraping; difficulties in extracting html template links, search engine URL, and finding the exact coordinates of these links
- Working with new libraries such as pyautogui, mouse events, etc
Accomplishments that we're proud of 💪:
- Successfully transferred ML model on Jupyter notebook to Google Cloud Platform (creating bucket, deploying functions, maintaining servers)
- Coding a responsive GUI (duck figure) based on mouse and keyboard detection from scratch, using OOP
- Working effectively as a team with a project that demanded a variety of sections (UI / UX Design, Building ML model, Developing python scripts)
What we learned 🧠:
- Implementing and deploying ML model on the cloud
- Working with new python tools, like selenium, to extract detailed web page information
What's next for Detective Duck - Phishing Detection Application 💼:
- Finetuning machine learning model
- Adding more features to our GUI
- Making our coordinate detection more accurate
Built With
- google-cloud
- procreate
- pyautogui
- python
- scikit-learn
- selenium
- tkinter
Log in or sign up for Devpost to join the conversation.