wehub-resource-sync
77 lines
2.5 KiB
Markdown
77 lines
2.5 KiB
Markdown
# 🎬 Video RAG with Gemini
|
|
|
|
A Streamlit demo that allows you to upload videos and chat with them using Google's Gemini AI with multimodal capabilities.
|
|
|
|
## Features
|
|
|
|
- 📹 **Video Upload**: Support for multiple video formats (MP4, AVI, MOV, MKV, WEBM)
|
|
- 🤖 **AI-Powered Chat**: Ask questions about your video content using Gemini's advanced video understanding
|
|
- 💬 **Interactive Interface**: Clean chat interface with streaming responses
|
|
- 🔄 **Session Management**: Maintain chat history and video context
|
|
- ⚡ **Real-time Processing**: Upload and process videos with progress feedback
|
|
|
|
## Setup
|
|
|
|
1. **Install Dependencies**
|
|
```bash
|
|
pip install -r requirements.txt
|
|
```
|
|
|
|
2. **Get Gemini API Key**
|
|
- Visit [Google AI Studio](https://aistudio.google.com/app/apikey)
|
|
- Create a new API key
|
|
- Keep it secure - you'll enter it in the app
|
|
|
|
3. **Run the Application**
|
|
```bash
|
|
streamlit run app.py
|
|
```
|
|
|
|
## Usage
|
|
|
|
1. **Enter API Key**: Input your Gemini API key in the sidebar
|
|
2. **Upload Video**: Choose a video file (supported formats listed above)
|
|
3. **Wait for Processing**: The video will be uploaded and processed by Gemini
|
|
4. **Start Chatting**: Ask questions about your video content!
|
|
|
|
## Example Questions
|
|
|
|
- "What is happening in this video?"
|
|
- "Summarize the main events"
|
|
- "Who are the people in this video?"
|
|
- "What objects can you see?"
|
|
- "Describe the setting and environment"
|
|
- "What actions are taking place?"
|
|
|
|
## Technical Details
|
|
|
|
- **Video Processing**: Uses Gemini's File API for video upload and processing
|
|
- **Multimodal AI**: Combines video understanding with natural language processing
|
|
- **File Size Limits**: Large files (>100MB) may take longer to process
|
|
- **Supported Formats**: MP4, AVI, MOV, MKV, WEBM
|
|
|
|
## Limitations
|
|
|
|
- Video processing time depends on file size and complexity
|
|
- Large files may fail to upload or process
|
|
- API rate limits may apply based on your Gemini API plan
|
|
- Some video formats may not be supported
|
|
|
|
## Troubleshooting
|
|
|
|
- **Upload Fails**: Check video format and file size
|
|
- **Processing Stuck**: Wait a few minutes, large files take time
|
|
- **API Errors**: Verify your API key is correct and has sufficient quota
|
|
- **No Response**: Try refreshing the page and re-uploading the video
|
|
|
|
## Built With
|
|
|
|
- [Streamlit](https://streamlit.io/) - Web app framework
|
|
- [Google Gemini API](https://ai.google.dev/gemini-api) - Multimodal AI capabilities
|
|
- [Python](https://python.org/) - Backend processing
|
|
|
|
---
|
|
|
|
*Part of the AI Engineering Hub - Building practical AI applications*
|
|
|