Vicapp

Vicapp is a smart video captioning tool built for the AMD Hackathon. It takes any video and automatically generates four different styles of captions using Google's Gemini AI.The app comes with three pre-loaded demo videos an autumn traffic scene, a curious kitten, and an office setting each already captioned and ready to explore. But the real power is the custom feature: you can paste any YouTube link and the app will download the video, analyze it frame by frame using Gemini's vision AI, and generate four unique captions in real-time. The four captioning styles are: Formal (professional and factual), Sarcastic (dry, witty observations), Humorous Tech (funny but with programming and tech references), and Humorous Non-Tech (everyday relatable humor).Under the hood, the app runs a React frontend with a dark, modern interface and a Python FastAPI backend that handles video downloading via yt-dlp, uploads the video to Google's cloud, and queries Gemini's vision model for each caption style. The whole pipeline from pasting a link to seeing captions takes about 30 to 60 seconds. The project is designed as a web app making it flexible for demo purposes or larger-scale use. It's a practical demonstration of how modern AI can understand and describe video content in multiple creative ways.