Velocity

Class: VelocityBlockV1

Source: inference.core.workflows.core_steps.analytics.velocity.v1.VelocityBlockV1

The VelocityBlock computes the velocity and speed of objects tracked across video frames. It includes options to smooth the velocity and speed measurements over time and to convert units from pixels per second to meters per second. It requires detections from Byte Track with unique tracker_id assigned to each object, which persists between frames. The velocities are calculated based on the displacement of object centers over time.

Note: due to perspective and camera distortions calculated velocity will be different depending on object position in relation to the camera.

Type identifier

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

Properties

Name	Type	Description	Refs
name	str	Enter a unique identifier for this step..	❌
smoothing_alpha	float	Smoothing factor (alpha) for exponential moving average (0 < alpha <= 1). Lower alpha means more smoothing..	✅
pixels_per_meter	float	Conversion from pixels to meters. Velocity will be converted to meters per second using this value..	✅

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 Velocity in version v1.

• inputs: Detections Stabilizer, Path Deviation, Time in Zone, Path Deviation, Perspective Correction, Gaze Detection, Detections Transformation, Instance Segmentation Model, Byte Tracker, Byte Tracker, PTZ Tracking (ONVIF).md), Byte Tracker, Cosine Similarity, Object Detection Model, Velocity, Detections Classes Replacement, YOLO-World Model, Detections Merge, Time in Zone, VLM as Detector, Template Matching, Instance Segmentation Model, Dynamic Zone, Dynamic Crop, Time in Zone, Line Counter, Detections Consensus, Moondream2, Overlap Filter, Google Vision OCR, Segment Anything 2 Model, Detections Stitch, Bounding Rectangle, Detection Offset, Camera Focus, Object Detection Model, Identify Changes, Detections Filter, VLM as Detector
• outputs: Detections Stabilizer, Roboflow Dataset Upload, Circle Visualization, Path Deviation, Time in Zone, Path Deviation, Perspective Correction, Color Visualization, Detections Transformation, Mask Visualization, Byte Tracker, Byte Tracker, Florence-2 Model, Blur Visualization, PTZ Tracking (ONVIF).md), Pixelate Visualization, Byte Tracker, Velocity, Dot Visualization, Stability AI Inpainting, Roboflow Dataset Upload, Detections Classes Replacement, Detections Merge, Icon Visualization, Time in Zone, Roboflow Custom Metadata, Model Monitoring Inference Aggregator, Polygon Visualization, Dynamic Zone, Dynamic Crop, Crop Visualization, Stitch OCR Detections, Time in Zone, Ellipse Visualization, Detections Consensus, Line Counter, Trace Visualization, Bounding Box Visualization, Distance Measurement, Segment Anything 2 Model, Overlap Filter, Label Visualization, Line Counter, Corner Visualization, Detections Stitch, Bounding Rectangle, Size Measurement, Detection Offset, Model Comparison Visualization, Background Color Visualization, Detections Filter, Florence-2 Model, Halo Visualization, Triangle Visualization

Input and Output Bindings

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

Bindings

• input

  • image (image): not available.
  • detections (Union[instance_segmentation_prediction, object_detection_prediction]): Model predictions to calculate the velocity for..
  • smoothing_alpha (float): Smoothing factor (alpha) for exponential moving average (0 < alpha <= 1). Lower alpha means more smoothing..
  • pixels_per_meter (float): Conversion from pixels to meters. Velocity will be converted to meters per second using this value..

• output

  • velocity_detections (Union[object_detection_prediction, instance_segmentation_prediction]): Prediction with detected bounding boxes in form of sv.Detections(...) object if object_detection_prediction or Prediction with detected bounding boxes and segmentation masks in form of sv.Detections(...) object if instance_segmentation_prediction.

Example JSON definition of step Velocity in version v1

{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/velocity@v1",
    "image": "<block_does_not_provide_example>",
    "detections": "$steps.object_detection_model.predictions",
    "smoothing_alpha": 0.5,
    "pixels_per_meter": 0.01
}