StreamRAG: Video Search & Chat with Your Videos 117

StreamRAG, a video search and streaming proxy tool, StreamRAG enables you to chat with your video library and watch video streams within ChatGPT.

StreamRAG is able to locate relevant video moments matching your needs from hundreds of hours of video content and instantly return a video clip, meaning it can search any moment within video content.

StreamRAG can rapidly browse through large amounts of stored video materials, find video segments containing such content or topics, and present these segments to you, so you can directly watch the portions of video that are relevant to what you searched for.

What does StreamRAG do?

StreamRAG allows users to upload videos, create video collections, and search across these videos to get real-time video responses or edits. In addition, users can publish their video collections to the ChatGPT Store, so that others can search and use them.

• Video Library Creation: Upload multiple videos to create a video library or collection.

• Video Search and Response: Search across these videos and instantly receive real-time video responses or compiled results.

• GPTs Publishing: Publish your searchable collections on ChatGPT's GPT Store, which means you can share your rich video content with a broader audience, or keep it private for personal use or within selected groups.

• Text Answer Summarization (RAG): Receive a summarized summary text answer (RAG).

• Video Key Insights: Get key insights from specific videos, such as "key points of episode 31".

• Open Source and Easy to Use: StreamRAG is open source, ensuring it is accessible and modifiable to meet all your needs. Even if you are unfamiliar with ai and video databases, StreamRAG is designed to be user-friendly and easy to implement.
  

How to use StreamRAG?

1. Get an API Key: Sign up on the VideoDB console (the first 50 uploads are free, no credit card required).

2. Set up VIDEO_DB_API_KEY: Enter your key in the file. .env

3. Install dependencies: Run this in the terminal.

4. Upload the collection to VideoDB: Add the link in. upload.py

5. Run locally: Use to start the flask server. python app.py