Brandon

  • Mar 4, 2023
  • Joined Feb 11, 2019
  • 0 best answers
  • In Share Your OAK Setups and Win Prizes!

    Raspberry Pi/Oak D Lite Depth Camera
    Setup for a portable, battery powered digital depth camera.
    Generates RGBD depth images compatible with the Looking Glass Portrait 3D display.
    Here is a link to the project at Prusa Printables with the full BOM and 3D printable mount.
    [https://www.printables.com/model/196422-raspberry-pi-4b-oak-d-lite-portable-3d-camera]


    The built in touchscreen display:

    And the output as shown on the Looking Glass Portrait:

    Video of the display in action, its hard to show how good this looks IRL.

  • In Share Your OAK Setups and Win Prizes!

    Tradeshow model: Automatic stockpile detection and volume computation

    We have built a showcase to give a hands-on experience for our industry and construction customers to play with gravel and sand and have the computer automatically detect stockpiles, compute area/volume and speak the results out loud.

    We are offering a similar service on large scale with drone flights -> UAV data acquisition, processing, offering data on a platform, automatically detect stockpiles in data and compute volumes in 3D point cloud data.

  • In Whitebalance problem

    Thanks for help. I really appreciate it. It would really help us

  • In Whitebalance problem

    Hi oto313 ,
    Let me ask the firmware team if there is any quick way for us to implement it. I assume you could also do this on your end using a custom CV node (kornia might have a function for that already), but it's not as simple.
    Thanks, Erik

    • In Changing Zoom at Runtime

      Hello hussain_allawati , I am not sure if ImageManip supports NV12 frames by default. Could you try moving to image_manip_refactor branch of depthai-python, running python3 examples/install_requirements.py (to get the latest version of depthai), and retrying the same script?
      Thanks, Erik

      • In Changing Zoom at Runtime

        Brandon Hello Brandon,
        I am having the same issue as @atb4c and would like to resize a 1080p NV12 Image into a 300 by 300 RGB Image.
        I want to capture STILL images at 1080p, and then resize them too 300 by 300 to feed them to a NN.
        My code is as follows:

        import cv2
import depthai as dai

xxx = cv2.imread("blue.png")

# Create the pipeline
pipeline = dai.Pipeline()

# Create Camera node and give its properties
camRGB = pipeline.create(dai.node.ColorCamera)
camRGB.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
camRGB.setStillSize(1080, 1080)
camRGB.setPreviewSize(1080, 1080)
camRGB.setVideoSize(1080, 1080)
# camRGB.setInterleaved(False)
camRGB.setColorOrder(dai.ColorCameraProperties.ColorOrder.RGB)


# Create Image manipulation node, to resize image from 1080p to 300by300
manip = pipeline.create(dai.node.ImageManip)
manip.initialConfig.setResize(300,300)
manip.initialConfig.setFrameType(dai.ImgFrame.Type.RGB888p)

# Create output node for still images
outStillRGB = pipeline.create(dai.node.XLinkOut)
outStillRGB.setStreamName("rgbStill")

# Create input control node to acquire capture command
xinCaptureCommand = pipeline.create(dai.node.XLinkIn)
xinCaptureCommand.setStreamName("capture")

# Link output of encoder to input of xlinkout to send to deivce
camRGB.still.link(manip.inputImage)
manip.out.link(outStillRGB.input)

# Link output of xinCaptureCommand to camera input control
xinCaptureCommand.out.link(camRGB.inputControl)

# Connect to device and start the pipeline
with dai.Device(pipeline) as device:
    # Create output queue that will get RGB frame (Output from device, and input to host)
    stillQueue = device.getOutputQueue(name="rgbStill")


    # Create input queue to device, that receives capture command
    captureInputQueue = device.getInputQueue("capture")

    cv2.imshow("xxx",xxx)
    while True:
       
       
       
        stillFrame = stillQueue.tryGet()
        if stillFrame is not None:
            frame = stillFrame.getCvFrame()
            #frame = cv2.imdecode(stillFrame.getData(), cv2.IMREAD_UNCHANGED)
            cv2.imshow("frame", frame)

        # Send capture command from host to device
        key = cv2.waitKey(1)
        if key == ord("q"):
            break
           
        elif key == ord('c'):
            ctrl = dai.CameraControl()
            ctrl.setCaptureStill(True)
            captureInputQueue.send(ctrl)
            print("captured")

        The problems is that the output I get is a grayscale image.
        Any thoughts on how to resolve the issue?

        • In Running pipeline on Docker container

          Hi @Brandon ,

          I have been using the depthai sample code to run natively on both AMD64 and ARM64 platforms successfully against an OAK camera. Although if I try to run through the same code samples from a docker container, I consistently get the X_LINK_DEVICE_NOT_FOUND error as seen below.

          root@dsodatdev7-desktop:/home/dso/depthai-python/examples/MobileNet# python3 rgb_mobilenet.py
          Traceback (most recent call last):
          File "rgb_mobilenet.py", line 56, in <module>
          with dai.Device(pipeline) as device:
          RuntimeError: Failed to find device after booting, error message: X_LINK_DEVICE_NOT_FOUND

          I am following the recipe as described previously above in the docker run command. Is there any light you can shed on this error.

          Thank you

          Joseph E. Martinez

          • In It Works! Working Prototype of Commute Guardian.

            Hello Jonathan , for measuring light brightness I would say you would first have to do some measurements to get reference points - eg. brightness X corresponds to Y pixel value. You could also set lower exposure time/sensitivity so not all light sources would max out pixel value (255) and you would only perceive light sources.
            For height, you would need to know the location and orientation (angle) of the camera. You could get headlight spatial coordinates with triangulation, like Brandon mentioned, demo here. After getting the distance (from disparity - red line in the demo GIF), you can calculate spatial coordinates and from that, you could calculate what's the headlight height on the car. Thoughts on such approach?
            Thanks, Erik

            • In It Works! Working Prototype of Commute Guardian.

              Brandon
              Can you point me in the direction of a minimal example? Much appreciated, and thanks in advance.

                • In OAK-D PoE Auto Exposure Failures

                  To follow up here, the other OAK-D PoE units I received more recently do not exhibit this issue. Luxonis-Alex in the discord discussion suggested that it might be a hardware issue, so I will do an RMA for the unit. Perhaps a test could be added to your QA process to ensure that the exposure works as expected in a few different lighting conditions for every unit, so this issue would be caught before shipment if it happens again.

                  • In Raw image from rgb camera that is demosaiced

                    Thank you Alex! I have my setup working so I'm happy to wait. From an efficiency point of view though, having the spi do the heavy lifting makes more since. Looking forward to the new output type 😀

                    T

                  • In It Works! Working Prototype of Commute Guardian.

                    For what it's worth, I recently came across https://ride.vision/ as well.

                    re. cars, I've noticed that recent model years (<5 yrs old?) seem to have headlights that are considerably brighter than older model years, to the point that they virtually blind oncoming traffic. Furthermore, SUVs and pickup trucks seem to have their headlights at roughly the same height as a sedan driver's eyes, so it's a double whammy.

                    Q - is it possible to measure light brightness?
                    Q - is it possible to estimate headlight height of oncoming traffic? (I have some ideas here but would love to hear suggestions).

                    • In OAK-D for calculating distance between people

                      erik checking it asap. Thanks a lot Erik. Especially, I'd like to "retrieve" and parse coordinates as text over udp or something like this. Ideally in openframeworks based exe, or max msp. I'll let you know.

                      • In Using DepthAI and multi-threading in C++

                        themarpe
                        My application is big, I tried to build a smaller application to reproduce the problem but without success so far. I will share it as soon as I have it

                        • In How to assign static IPs to POE devices
                        • In Subpixel doesn't work when running gen2-collecting-training-data
                        • In Subpixel doesn't work when running gen2-collecting-training-data

                          Hello nickjrz ,
                          so disparity is being colorized on this line. By default, disparity has values [0..95], so within a byte. After you add subpixel mode, however, disparity takes up 2 bytes, as it's values are [0..3040], more info here. So if you want to fit 0..3040 values into 0..255, you could divide the whole frame by 11,921568627 and convert these values to uint8.
                          Thanks, Erik

                            • In module 'depthai' has no attribute 'create_pipeline'

                              I am able to run other sample codes successfully. I noticed the sample codes use depthai.pipeline() to create pipeline.

                              I read this issue on github and user mentioned to call create_pipeline before request_af_trigger.

                            • In OAK-D IMU sequence number and timestamps don't line up

                              Shaadee
                              Hello.
                              Gyro and accelerometer sequence numbers are different because they are running at different sampling rates internally, accelerometer at 500 hz, gyro at 400.
                              You don't need to sync them on host side, when multiple outputs are enabled for the IMU node they are automatically synced.

                              For syncing with Left/Right cameras the best way is using timestamp information.

                              • In OpenCV OAK Course?

                                nalzok actually the course was already updated to the gen2 of our library (API). There might not be everything there (as we have improved our library a lot) but it's not outdated.
                                Thanks, Erik