What does it do?
The Image – Embedding connector converts images into high-dimensional vector representations using advanced neural network models. This component generates vector embeddings from images, converting visual content into numerical representations that capture semantic meaning and visual features. These embeddings enable similarity search, clustering, classification, and other AI-powered image analysis tasks.

Inputs
With the Image – Embedding connector, you can:
- Transform images into vectors for use in vector databases or search engines
- Enable image similarity search and content-based retrieval
- Prepare image data for downstream machine learning or AI workflows
- Cluster images by visual similarity for organization or deduplication
- Feed image vectors into AI models for classification or anomaly detection
Inputs and Outputs
Inputs
- Images – Image files to convert to embeddings
- Documents – Document objects containing images
Outputs
- Vectors – Generated vector embeddings
- Documents – Original documents with embeddings attached
How do I use it?
To use the Image – Embedding connector in your workflow:
- Add the Image – Embedding Connector
- Insert the node into your pipeline where you want to generate embeddings from images
- Connect Input
- Connect the input lane (image or documents) to your image source
- This could be a file dropper, file system, image parser, or any image source
- Configure Parameters
- Adjust the embedding model and processing options as needed
- Configure image preprocessing settings based on your requirements
- Connect Output
- The connector outputs generated image embeddings
- Send these to downstream nodes for similarity search, clustering, or further analysis
Configuration
Model Settings
| Parameter | Description | Default/Options |
|---|---|---|
| Model | Image embedding model to use | Default: “clip-vit-base-patch32” Available: CLIP, ResNet, EfficientNet |
| Dimensions | Vector dimensions | Default: 512 (model dependent) |
| Batch Size | Number of images to embed at once | Default: 16 (affects memory usage) |
| Normalize | Normalize vector lengths | Default: true (improves similarity calculations) |
Available Model Options
- Custom model – User-defined embedding engine, not configured in this view
- Google – 16×16 – Fast, accurate, general-purpose embeddings
- OpenAI – 16×16 – Good performance with lower memory usage
- OpenAI – 32×32 – Lower performance, better image recognition
Image Processing
| Parameter | Description | Default/Options |
|---|---|---|
| Resize | Resize images before embedding | Default: true (ensures consistent input size) |
| Target Size | Target image dimensions | Default: [224, 224] (width and height in pixels) |
| Center Crop | Apply center cropping | Default: true (maintains aspect ratio) |
| Color Mode | Color processing mode | Default: “RGB” (Options: RGB, grayscale) |
Advanced Settings
| Parameter | Description | Default/Options |
|---|---|---|
| Cache | Cache embeddings for reuse | Default: true (improves performance for repeated images) |
| Device | Processing device | Default: “auto” (Options: auto, cpu, cuda) |
| Precision | Computation precision | Default: “float32” (Options: float32, float16, bfloat16) |
Example Use Cases
- Enable reverse image search or “find similar images” features
- Cluster images by visual similarity for organization or deduplication
- Feed image vectors into AI models for classification or anomaly detection
- Build content-based image recommendation systems
- Create visual search engines for e-commerce or media applications
- Perform image quality assessment and analysis
- Enable automated image tagging and categorization
Best Practices
Model Selection
- Use CLIP models for general-purpose image embeddings
- Use ResNet models for traditional computer vision tasks
- Use EfficientNet for resource-constrained environments
Image Preparation
- Ensure consistent image sizes through resizing
- Consider image quality and resolution
- Use center cropping to maintain important visual elements
- Preprocess images to remove noise or irrelevant content
Performance Optimization
- Adjust batch size based on available memory
- Use GPU acceleration when available
- Enable caching for repeated processing of the same images
- Use lower precision (float16) for faster processing with minimal quality loss
Troubleshooting
Processing Problems
- Out of memory errors – Reduce batch size or image dimensions
- Slow processing – Enable GPU acceleration or reduce image size
- Poor embedding quality – Try different models or image preprocessing
Compatibility Issues
- Model loading errors – Verify model availability and compatibility
- Device errors – Check CUDA installation for GPU acceleration
- Format errors – Ensure images are in supported formats (JPEG, PNG, etc.)
In summary:
The Image – Embedding connector transforms images into vector embeddings, enabling powerful AI-driven image search, clustering, and analysis workflows. With comprehensive configuration options for models, preprocessing, and performance optimization, it provides a flexible foundation for image-based AI applications.
