Model Monitoring Inference Aggregator

Class: ModelMonitoringInferenceAggregatorBlockV1

Source: inference.core.workflows.core_steps.sinks.roboflow.model_monitoring_inference_aggregator.v1.ModelMonitoringInferenceAggregatorBlockV1

This block 📊 transforms inference data reporting to a whole new level by periodically aggregating and sending a curated sample of predictions to Roboflow Model Monitoring.

✨ Key Features

• Effortless Aggregation: Collects and organizes predictions in-memory, ensuring only the most relevant and confident predictions are reported.

• Customizable Reporting Intervals: Choose how frequently (in seconds) data should be sent—ensuring optimal balance between granularity and resource efficiency.

• Debug-Friendly Mode: Fine-tune operations by enabling or disabling asynchronous background execution.

🔍 Why Use This Block?

This block is a game-changer for projects relying on video processing in Workflows. With its aggregation process, it identifies the most confident predictions across classes and sends them at regular intervals in small messages to Roboflow backend - ensuring that video processing performance is impacted to the least extent.

Perfect for:

• Monitoring production line performance in real-time 🏭.

• Debugging and validating your model’s performance over time ⏱️.

• Providing actionable insights from inference workflows with minimal overhead 🔧.

🚨 Limitations

• The block is should not be relied on when running Workflow in inference server or via HTTP request to Roboflow hosted platform, as the internal state is not persisted in a memory that would be accessible for all requests to the server, causing aggregation to only have a scope of single request. We will solve that problem in future releases if proven to be serious limitation for clients.

• This block do not have ability to separate aggregations for multiple videos processed by InferencePipeline - effectively aggregating data for all video feeds connected to single process running InferencePipeline.

Type identifier

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

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
frequency	int	Frequency of reporting (in seconds). For example, if 5 is provided, the block will report an aggregated sample of predictions every 5 seconds..	✅
unique_aggregator_key	str	Unique key used internally to track the session of inference results reporting. Must be unique for each step in your Workflow..	❌
fire_and_forget	bool	Boolean flag to run the block asynchronously (True) for faster workflows or synchronously (False) for debugging and error handling..	✅

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 Model Monitoring Inference Aggregator in version v1.

• inputs: Bounding Rectangle, Identify Outliers, Multi-Label Classification Model, Single-Label Classification Model, Webhook Sink, Dynamic Crop, Clip Comparison, Google Vision OCR, Twilio SMS Notification, Segment Anything 2 Model, Detection Offset, Model Monitoring Inference Aggregator, Florence-2 Model, LMM For Classification, Roboflow Dataset Upload, Identify Changes, CogVLM, OCR Model, Template Matching, VLM as Detector, JSON Parser, SIFT Comparison, Multi-Label Classification Model, Path Deviation, OpenAI, Stitch OCR Detections, Velocity, Detections Stitch, OpenAI, Detections Stabilizer, Path Deviation, VLM as Classifier, Time in Zone, Keypoint Detection Model, Perspective Correction, Moondream2, SIFT Comparison, Slack Notification, Anthropic Claude, Email Notification, LMM, Llama 3.2 Vision, Instance Segmentation Model, CSV Formatter, VLM as Detector, Time in Zone, Single-Label Classification Model, Byte Tracker, Florence-2 Model, Detections Filter, Detections Transformation, YOLO-World Model, Detections Consensus, Keypoint Detection Model, Google Gemini, Detections Merge, Local File Sink, Dynamic Zone, Instance Segmentation Model, Roboflow Custom Metadata, VLM as Classifier, Detections Classes Replacement, Gaze Detection, Object Detection Model, Roboflow Dataset Upload, Byte Tracker, Byte Tracker, Line Counter, Object Detection Model
• outputs: Multi-Label Classification Model, Single-Label Classification Model, Classification Label Visualization, Webhook Sink, Background Color Visualization, Dynamic Crop, Cache Get, Mask Visualization, Clip Comparison, Twilio SMS Notification, Google Vision OCR, Segment Anything 2 Model, Model Monitoring Inference Aggregator, Stability AI Image Generation, LMM For Classification, Florence-2 Model, Image Blur, Cache Set, Roboflow Dataset Upload, CogVLM, Line Counter, Circle Visualization, Crop Visualization, Template Matching, Multi-Label Classification Model, Path Deviation, OpenAI, Detections Stitch, OpenAI, Pixel Color Count, Path Deviation, Line Counter, Time in Zone, Model Comparison Visualization, Bounding Box Visualization, Keypoint Detection Model, SIFT Comparison, Perspective Correction, Color Visualization, Slack Notification, Ellipse Visualization, Reference Path Visualization, Blur Visualization, Pixelate Visualization, Anthropic Claude, Email Notification, LMM, Llama 3.2 Vision, Instance Segmentation Model, Keypoint Visualization, Time in Zone, Single-Label Classification Model, Florence-2 Model, YOLO-World Model, Trace Visualization, Image Threshold, Polygon Visualization, Triangle Visualization, Detections Consensus, Keypoint Detection Model, Stability AI Inpainting, Halo Visualization, Dot Visualization, Polygon Zone Visualization, Google Gemini, CLIP Embedding Model, Local File Sink, Size Measurement, Instance Segmentation Model, Roboflow Custom Metadata, Gaze Detection, Object Detection Model, Corner Visualization, Line Counter Visualization, Roboflow Dataset Upload, Image Preprocessing, Label Visualization, Distance Measurement, Object Detection Model

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Model Monitoring Inference Aggregator in version v1 has.

Bindings

• input

  • predictions (Union[keypoint_detection_prediction, classification_prediction, object_detection_prediction, instance_segmentation_prediction]): Model predictions to report to Roboflow Model Monitoring..
  • model_id (roboflow_model_id): Model ID to report to Roboflow Model Monitoring..
  • frequency (string): Frequency of reporting (in seconds). For example, if 5 is provided, the block will report an aggregated sample of predictions every 5 seconds..
  • fire_and_forget (boolean): Boolean flag to run the block asynchronously (True) for faster workflows or synchronously (False) for debugging and error handling..

• output

  • error_status (boolean): Boolean flag.
  • message (string): String value.

Example JSON definition of step Model Monitoring Inference Aggregator in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/model_monitoring_inference_aggregator@v1",
    "predictions": "$steps.my_step.predictions",
    "model_id": "my_project/3",
    "frequency": "3",
    "unique_aggregator_key": "session-1v73kdhfse",
    "fire_and_forget": true
}