DETECTOR AI

Inspiration Misinformation is one of the biggest challenges today. With fake images, manipulated media, and misleading articles spreading rapidly, it becomes harder for people to distinguish fact from fiction. Our inspiration was to build a tool that helps users verify images and text in real time — and even learn from it through a quiz-based game. What it does DETECTOR AI is a Flask-based web application that: Verifies text and images using Gemini AI. Classifies inputs as Real, Fake, or Unknown with confidence scores. Explains why the decision was made in clear language. Provides an interactive quiz game that generates questions from the analyzed image/text so users can learn and test themselves. How we built it Backend: Flask (Python) Frontend: HTML, CSS, JavaScript (with animated quiz UI) Database/Session: Flask sessions for storing quiz progress and streaks AI/ML: Google Gemini API for text/image analysis & quiz generation Version control: Git + GitHub (with branches like gemini-image-verification) Other tools: CORS, requests, JSON parsing, base64 encoding Challenges we ran into Parsing Gemini responses into clean JSON (sometimes wrapped in json … ). Ensuring the image verification didn’t break when API returned unexpected results. Designing the quiz so options “fall down” with smooth animations. Managing sessions in Flask to track user progress. Accomplishments we’re proud of Successfully integrated text + image verification in one platform. Built an educational quiz game that makes learning about misinformation engaging. Overcame JSON parsing issues with Gemini outputs. Clean Git branching workflow (main, gemini-image-verification, quiz-game). What we learned How to integrate Gemini APIs with Flask. The importance of strict JSON parsing when working with AI APIs. Frontend animation tricks (CSS keyframes, JS timing) for gamified experiences. Better teamwork and version control with Git branches. What’s next Add leaderboards for the quiz game. Enable multi-user login with Firebase/Auth. Deploy on Render/Heroku with a persistent database. Expand support to detect deepfakes and videos

Built With

• api
• base64
• flask
• gemini
• github
• google
• html/css/javascript
• json
• python