Identify Changes

Class: IdentifyChangesBlockV1

Source: inference.core.workflows.core_steps.sampling.identify_changes.v1.IdentifyChangesBlockV1

Identify changes compared to prior data via embeddings.

This block accepts an embedding and compares it to a prior average and standard deviation for the rate of change. When things change faster or slower than they have in the past, the block will flag the data as an outlier.

Type identifier

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

Properties

Name	Type	Description	Refs
name	str	Unique name of step in workflows.	❌
strategy	str	The change identification algorithm to use..	❌
threshold_percentile	float	The desired sensitivity. A higher value will result in more data points being classified as outliers..	✅
warmup	int	The number of data points to use for the initial average calculation. No outliers are identified during this period..	✅
smoothing_factor	float	The smoothing factor for the EMA algorithm. The default of 0.25 means the most recent data point will carry 25% weight in the average. Higher values will make the average more responsive to recent data points..	✅
window_size	int	The number of data points to consider in the sliding window algorithm..	✅

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 Identify Changes in version v1.

• inputs: Identify Outliers, Image Contours, Template Matching, Perspective Correction, Line Counter, Detections Consensus, Distance Measurement, Perception Encoder Embedding Model, Line Counter, Pixel Color Count, Clip Comparison, SIFT Comparison, CLIP Embedding Model, Identify Changes, SIFT Comparison
• outputs: Keypoint Visualization, Keypoint Detection Model, Google Gemini, Detections Stabilizer, Circle Visualization, Perspective Correction, Color Visualization, Gaze Detection, Instance Segmentation Model, Reference Path Visualization, Stitch Images, PTZ Tracking (ONVIF).md), Blur Visualization, Keypoint Detection Model, Relative Static Crop, Cosine Similarity, Stability AI Inpainting, SIFT Comparison, Icon Visualization, Time in Zone, Roboflow Custom Metadata, Polygon Zone Visualization, Identify Outliers, Instance Segmentation Model, Template Matching, Stability AI Image Generation, Dynamic Zone, Dynamic Crop, Crop Visualization, Time in Zone, Detections Consensus, Single-Label Classification Model, Camera Calibration, Segment Anything 2 Model, Object Detection Model, Twilio SMS Notification, Model Comparison Visualization, Llama 3.2 Vision, Triangle Visualization, Line Counter Visualization, Multi-Label Classification Model, Email Notification, Roboflow Dataset Upload, Time in Zone, Image Slicer, Mask Visualization, Byte Tracker, Byte Tracker, Single-Label Classification Model, Pixelate Visualization, Webhook Sink, Byte Tracker, Object Detection Model, Slack Notification, Velocity, Dot Visualization, Image Slicer, Roboflow Dataset Upload, YOLO-World Model, Classification Label Visualization, Model Monitoring Inference Aggregator, Polygon Visualization, OpenAI, Ellipse Visualization, Trace Visualization, Stability AI Outpainting, Bounding Box Visualization, Distance Measurement, Multi-Label Classification Model, Label Visualization, Corner Visualization, Detections Stitch, Background Color Visualization, Identify Changes, Halo Visualization, OpenAI, Anthropic Claude

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Identify Changes in version v1 has.

Bindings

• input

  • embedding (embedding): Embedding of the current data..
  • threshold_percentile (float_zero_to_one): The desired sensitivity. A higher value will result in more data points being classified as outliers..
  • warmup (integer): The number of data points to use for the initial average calculation. No outliers are identified during this period..
  • smoothing_factor (float_zero_to_one): The smoothing factor for the EMA algorithm. The default of 0.25 means the most recent data point will carry 25% weight in the average. Higher values will make the average more responsive to recent data points..
  • window_size (integer): The number of data points to consider in the sliding window algorithm..

• output

  • is_outlier (boolean): Boolean flag.
  • percentile (float_zero_to_one): float value in range [0.0, 1.0].
  • z_score (float): Float value.
  • average (embedding): A list of floating point numbers representing a vector embedding..
  • std (embedding): A list of floating point numbers representing a vector embedding..
  • warming_up (boolean): Boolean flag.

Example JSON definition of step Identify Changes in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/identify_changes@v1",
    "strategy": "Simple Moving Average (SMA)",
    "embedding": "$steps.clip.embedding",
    "threshold_percentile": "$inputs.sample_rate",
    "warmup": 100,
    "smoothing_factor": 0.1,
    "window_size": 5
}