SceneSniffer

Inspiration

The inspiration for SceneSniffer came from the common frustration of watching an interesting scene from a movie or TV show in a YouTube Short without knowing its source. The goal was to create a tool that could instantly identify the content, so viewers would never again struggle to find the source of an amazing scene they just watched.

What it does

SceneSniffer is a Chrome extension that identifies whether a YouTube Shorts video is from a movie or a TV series. When a clip is from a TV show, the extension provides specific details like the show name, season, episode number, and episode title. It works with a single click, analysing video metadata like the title, description, and thumbnail to provide accurate results.

How we built it

SceneSniffer is built as a Manifest V3 Chrome extension. The core of its functionality relies on an AI model to perform the analysis.

The architecture involves the following components:

• Frontend: A user-facing popup built with HTML (popup.html) serves as the interface. A content script ( content.js) interacts with the YouTube website.
• Backend Logic: The extension uses the Gemini API for its AI-powered analysis. It captures screenshots of the video, which are then sent to the Google Gemini API for identification.
• Dependencies: The project utilises Node.js packages, including @google/genai and @google/generative-ai to interface with the AI service, and axios for making HTTP requests. It uses Webpack as a development dependency.

The extension requires permissions to access storage, the active tab, and scripting capabilities to function on youtube.com and make requests to the generativelanguage.googleapis.com API endpoint.

Challenges we ran into

The main challenge we ran into was accurately identifying the content.

Accomplishments that we're proud of

The key accomplishments of this project are embodied in its core features:

• AI-Powered Analysis: The extension successfully leverages a sophisticated AI model to process various video data points for identification.
• High Accuracy: It achieves reliable results by combining multiple data sources with a weighted focus on the video's title and description.
• User-Friendly Interface: The project delivers a simple and intuitive design for seamless navigation and quick, one-click identification.

What's next for SceneSniffer

Beyond the initial release, the vision is to expand its capabilities. Currently, the extension only works on YouTube, and the next major goal is to make it functional on other video platforms as well.

Built With

• chrome
• gemini
• google
• html/css
• javascript