https://github.com/bigmb/mb_visualize
visualization package of embedidngs in docker/closed environment
https://github.com/bigmb/mb_visualize
Last synced: about 2 months ago
JSON representation
visualization package of embedidngs in docker/closed environment
- Host: GitHub
- URL: https://github.com/bigmb/mb_visualize
- Owner: bigmb
- Created: 2024-12-12T04:15:17.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2024-12-12T05:15:45.000Z (5 months ago)
- Last Synced: 2025-02-07T12:48:54.788Z (3 months ago)
- Language: Python
- Size: 12.7 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Embedding Visualization Tool
A Docker-based visualization tool for exploring embeddings with associated metadata and images. The tool provides an interactive scatter plot visualization with image display capabilities and metadata exploration.
## Features
- Interactive scatter plot visualization of embeddings
- Click on points to view associated images
- Select multiple points to view multiple images
- Display metadata (image path, date, taxcode) for selected points
- GUI application for easy file selection and viewer launch
- Real-time monitoring of new embedding files
- Docker containerization for easy deployment
## Project Structure
```
.
├── data/
│ ├── embeddings/ # Store embedding JSON files here
│ └── images/ # Store referenced images here
├── app.py # Main Dash application
├── gui.py # GUI for file selection and viewer launch
├── Dockerfile # Docker container configuration
├── docker-compose.yml # Docker Compose configuration
└── requirements.txt # Python dependencies
```
## Setup Instructions
1. Install Dependencies:
```bash
pip install -r requirements.txt
```
2. Docker Setup:
- Install Docker and Docker Compose
- Ensure Docker daemon is running
## Usage
1. Place your embedding files in the `data/embeddings` directory
- Files should be in JSON format with the following structure:
```json
{
"embeddings": [[x1, y1], [x2, y2], ...],
"metadata": [
{
"image": "path/to/image.jpg",
"date": "2023-11-20",
"taxcode": "ABC123"
},
...
]
}
```
2. Place corresponding images in the `data/images` directory
3. Launch the GUI:
```bash
python gui.py
```
4. Using the GUI:
- Select an embedding file from the list
- Click "Launch Viewer" to start the visualization
- The viewer will open in your default web browser
5. Interacting with the Visualization:
- Click individual points to view the corresponding image
- Click and drag to select multiple points
- Hover over points to see metadata
- Use the plot toolbar for zooming and panning
## Docker Management
Manual Docker commands (alternative to GUI):
```bash
# Build and start the container
docker-compose up -d
# Stop the container
docker-compose down
# View logs
docker-compose logs -f
```
## Notes
- The GUI automatically monitors the embeddings directory for new files
- Images are displayed in real-time when points are selected
- The Docker container mounts the local data directory, so files can be added without rebuilding
- The viewer runs on port 8050 by default
## Troubleshooting
1. If the viewer doesn't launch:
- Check if Docker is running
- Ensure port 8050 is available
- Check the Docker logs for errors
2. If images don't display:
- Verify image paths in the embedding file are correct
- Ensure images are present in the data/images directory
- Check file permissions
3. For GUI issues:
- Restart the GUI application
- Check if previous Docker containers are still running