YOLOv6 Model Not Detecting on OAK-D Lite

charb · 2025-03-19T04:36:07+00:00

Hi there, I have an OAK-D Lite camera and I want to use it to detect drones and their distance relative to the camera. I am using a YOLOv6-N model as it is lightweight and seems to be a pretty popular choice amongst OAK-D Lite projects. The script that I am running to detect drone's and their distance relative to the camera is:

from depthai_sdk
import OakCamera, ArgsParser
import argparse

# parse arguments
parser = argparse.ArgumentParser()
parser.add_argument("-conf", "--config", help="Trained YOLO json config path", default='models/v1/16shaves/droneDetection_v1.json', type=str)
args = ArgsParser.parseArgs(parser)

with OakCamera(args=args) as oak:
    color = oak.create_camera('color')
    nn = oak.create_nn(args['config'], color, nn_type='yolo', spatial=True)
    oak.visualize(nn, fps=True)
    oak.start(blocking=True)

This code outputs the following, the bottom-right of the bounding box lies the coveted Z-parameter - "0.9m":
(sorry for the low-res)

I want to extract the Z-parameter into a variable, and I am very unsure of how to do so. I've tried using many different examples from the documentation and other sources, but to no avail.

Any help would be appreciated!

jakaskerl · 2025-03-19T15:06:45+00:00

charb
You need to crate a callback function and pass in the nn.spatials. oak.add_callback(cb). https://docs-old.luxonis.com/projects/sdk/en/latest/samples/visualizer/SDK_visualizer_callback/

charb · 2025-03-19T16:03:06+00:00

I tried using the callback function example found in the link and that doesn't work with my model. I tried switching out DetectionPacket for DepthPacket and still no bounding boxes were drawn.

In my script, where is that Z-parameter being stored and how can I store that in a variable for handling? Here is a better image of the output:

charb · 2025-03-19T16:43:52+00:00

Nevermind. This script works:

from depthai_sdk import OakCamera, ArgsParser
import argparse

latest_z_value = None  # Store the most recent Z distance

def process_detections(packet):
    global latest_z_value

    if packet is None or not hasattr(packet, 'detections'):
        print("No detections found.")
        return  

    for det in packet.detections:
        if hasattr(det, 'img_detection') and hasattr(det.img_detection, 'spatialCoordinates'):
            latest_z_value = det.img_detection.spatialCoordinates.z / 1000
            print(f"Drone detected {latest_z_value:.2f}m away.")
        else:
            print("No spatial data available.")


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("-conf", "--config", help="Trained YOLO json config path", default='models/v1/16shaves/droneDetection_v1.json', type=str)
    args = ArgsParser.parseArgs(parser)

    with OakCamera(args=args) as oak:
        color = oak.create_camera('color')
        nn = oak.create_nn(args['config'], color, nn_type='yolo', spatial=True)
        nn.config_nn(resize_mode='stretch')
        
        oak.callback(nn, process_detections)
        oak.visualize(nn, fps=True)
        oak.start(blocking=True)

if __name__ == "__main__":
    main()