Dimension Collapse

Class: DimensionCollapseBlockV1

Source: inference.core.workflows.core_steps.fusion.dimension_collapse.v1.DimensionCollapseBlockV1

Takes multiple step outputs at data depth level n, concatenate them into list and reduce dimensionality to level n-1.

Useful in scenarios like: * aggregation of classification results for dynamically cropped images * aggregation of OCR results for dynamically cropped images

Type identifier

Use the following identifier in step "type" field: roboflow_core/dimension_collapse@v1to add the block as as step in your workflow.

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to Dimension Collapse in version v1.

• inputs: Crop Visualization, SIFT, Stitch OCR Detections, Blur Visualization, PTZ Tracking (ONVIF).md), Cache Set, Property Definition, Camera Focus, Mask Visualization, VLM as Classifier, Object Detection Model, Keypoint Detection Model, Image Preprocessing, Qwen2.5-VL, Object Detection Model, Bounding Rectangle, Cache Get, JSON Parser, SmolVLM2, LMM, Polygon Zone Visualization, QR Code Generator, Size Measurement, Florence-2 Model, Moondream2, Template Matching, Label Visualization, Pixel Color Count, Segment Anything 2 Model, Stability AI Image Generation, Keypoint Detection Model, Relative Static Crop, Slack Notification, Corner Visualization, Multi-Label Classification Model, Overlap Filter, Image Blur, Gaze Detection, Model Comparison Visualization, VLM as Detector, Time in Zone, Image Threshold, Roboflow Dataset Upload, CogVLM, Byte Tracker, Barcode Detection, Single-Label Classification Model, First Non Empty Or Default, Polygon Visualization, Stability AI Outpainting, Keypoint Visualization, Time in Zone, Grid Visualization, Delta Filter, OpenAI, Florence-2 Model, Image Slicer, Instance Segmentation Model, Twilio SMS Notification, Detections Stitch, Identify Changes, Continue If, Dynamic Crop, Anthropic Claude, QR Code Detection, Line Counter, Line Counter, Rate Limiter, LMM For Classification, Line Counter Visualization, Color Visualization, Image Contours, Image Convert Grayscale, Circle Visualization, Google Gemini, Data Aggregator, Cosine Similarity, Absolute Static Crop, Expression, Dynamic Zone, Multi-Label Classification Model, Environment Secrets Store, Byte Tracker, Detections Consensus, Stitch Images, Detections Filter, Trace Visualization, Roboflow Custom Metadata, OCR Model, Detections Transformation, Clip Comparison, YOLO-World Model, CLIP Embedding Model, Halo Visualization, Perspective Correction, Buffer, Stability AI Inpainting, Velocity, Webhook Sink, Distance Measurement, VLM as Detector, Perception Encoder Embedding Model, Triangle Visualization, Background Color Visualization, Detections Stabilizer, Byte Tracker, Path Deviation, Icon Visualization, SIFT Comparison, Pixelate Visualization, CSV Formatter, Instance Segmentation Model, Llama 3.2 Vision, Google Vision OCR, Reference Path Visualization, VLM as Classifier, Image Slicer, Identify Outliers, Depth Estimation, Roboflow Dataset Upload, Detection Offset, OpenAI, Classification Label Visualization, Dot Visualization, Email Notification, Local File Sink, OpenAI, Single-Label Classification Model, Bounding Box Visualization, Camera Calibration, Detections Classes Replacement, Ellipse Visualization, Detections Merge, Path Deviation, Model Monitoring Inference Aggregator, Dimension Collapse, SIFT Comparison, Dominant Color, Clip Comparison
• outputs: Anthropic Claude, Crop Visualization, Line Counter, Keypoint Detection Model, Line Counter, Triangle Visualization, LMM For Classification, Line Counter Visualization, Color Visualization, Cache Set, Corner Visualization, Path Deviation, Mask Visualization, Circle Visualization, Google Gemini, Webhook Sink, VLM as Detector, VLM as Classifier, Instance Segmentation Model, Object Detection Model, Time in Zone, Llama 3.2 Vision, Keypoint Detection Model, VLM as Classifier, Reference Path Visualization, Detections Consensus, Roboflow Dataset Upload, Roboflow Dataset Upload, Trace Visualization, OpenAI, Classification Label Visualization, Polygon Visualization, Keypoint Visualization, Time in Zone, Dot Visualization, Email Notification, Grid Visualization, Object Detection Model, Clip Comparison, Bounding Box Visualization, Polygon Zone Visualization, Ellipse Visualization, YOLO-World Model, Size Measurement, Halo Visualization, Florence-2 Model, OpenAI, Perspective Correction, Florence-2 Model, Buffer, Path Deviation, Instance Segmentation Model, Label Visualization, Clip Comparison, VLM as Detector

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Dimension Collapse in version v1 has.

Bindings

• input

  • data (*): Reference to step outputs at depth level n to be concatenated and moved into level n-1..

• output

  • output (list_of_values): List of values of any type.

Example JSON definition of step Dimension Collapse in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/dimension_collapse@v1",
    "data": "$steps.ocr_step.results"
}