detection/core/

[giscus[bot]]

detection/core/

A set of easy-to-use utilities that will come in handy in any computer vision project.

https://supervision.roboflow.com/detection/core/

[aberg2323]

Is there an example of how to use get_anchors_coordinates(anchor)?

[LinasKo]

It will be something like this:

detections = sv.Detections.from_inference(result)
center_points = detections. get_anchors_coordinates(sv.Position.CENTER)

The center points will have a shape (n, 2) - a set of x and y for every detected object in detections

[aberg2323]

You are seriously a life saver! Thank you!!

[aberg2323]

I was able to output the center points, however, I wish to save the center point that corresponds to each detection to the csv file that outputs the tracked detections. Right now, my code appends the csv file with all of the detections in the frame in which the object was detected, not the center point that specifically corresponds to that object. Do you have any suggestions on how I should fix this?

[LinasKo]

If you store the center point in the data dict of the Detections, the CSVSink should save it.

For example, in this Cookbook we are setting the "class_name" before saving the detections in a CSV.

While it's part of InferencePipeline callback and assumes you only have person detections at that point, the line we use is:

detections.data["class_name"] = np.array(["person" for _ in range(len(detections))])

Hope that helps as a starting point.