X_LINK_ERROR with ToF camera in a Jupyter notebook

spnsingh

Hello:

What is the best way to query a ToF camera (OAK-FFC ToF 33D) using DepthAI in a Jupyter Notebook?

When I take the tof_depth.py example code and run it directly in a Jupyter Notebook it gives communication errors on the dai.Device(pipeline).getOutputQueue(name="depth").get() call.

I followed the advise on a related issue with an 'rgb' camera and it seems to be connecting via USB3 when the python (not Jupyter notebook) is called.

Thanks!

PS. If it helps, please see the notebook is here. The full error is:

Cell In[10], line 4
1# Connect to device and start pipeline
2q = dai.Device(pipeline).getOutputQueue(name="depth") ---->
4 imgFrame = q.get() # blocking call, will wait until a new data has arrived
5depth_map = imgFrame.getFrame()
7# Colorize the depth frame to jet colormap RuntimeError: Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'depth' (X_LINK_ERROR)'

jakaskerl

Hi spnsingh
Does an RGB example work? Can you get preview stream from OV9782 camera?

Thanks,
Jaka

spnsingh

Hi Jaka,

At present, I just have a ToF 33D camera for the OAK-FFC-3P we have.
(I can order one and get back to you on this.)

Thanks,
Surya

jakaskerl

Hi spnsingh
Could you post the whole notebook please. I can test this locally to see if I can reproduce the issue.

Thanks,
Jaka

spnsingh

Hi jakaskerl

Gladly -- It is on Google Drive here.
(Sorry, the forum framework does not allow one to upload a Jupyter Notebook).

jakaskerl

Hi @spnsingh
Doesn't look like this is the full one though.

Thanks,
Jaka

spnsingh

Hi Jaka,

I have uploaded the Jupyter Notebook again (via Google Drive).

For simplicity, there is not much to it. It basically runs through the tof_depth.py example, but as a notebook.

What might be causing the 'Couldn't read data from stream: 'depth' (X_LINK_ERROR)' message?

Thanks,

Surya

jakaskerl

Hi @spnsingh
The way you have divided your blocks opens the code up for a bunch of errors. Please put the pipeline related code in its own cell.
You can put the rest in a separate cell but keep in mind that each time you use dai.Device(), the device will be opened and run.

Call the device with this code.

# Connect to device and start pipeline
with dai.Device(pipeline) as device:
    print('Connected cameras:', device.getConnectedCameraFeatures())
    q = device.getOutputQueue(name="depth")

    while True:
        imgFrame = q.get()  # blocking call, will wait until a new data has arrived
        depth_map = imgFrame.getFrame()

        # Colorize the depth frame to jet colormap
        depth_downscaled = depth_map[::4]
        non_zero_depth = depth_downscaled[depth_downscaled != 0] # Remove invalid depth values
        if len(non_zero_depth) == 0:
            min_depth, max_depth = 0, 0
        else:
            min_depth = np.percentile(non_zero_depth, 3)
            max_depth = np.percentile(non_zero_depth, 97)
        depth_colorized = np.interp(depth_map, (min_depth, max_depth), (0, 255)).astype(np.uint8)
        depth_colorized = cv2.applyColorMap(depth_colorized, cv2.COLORMAP_JET)

        cv2.imshow("Colorized depth", depth_colorized)

        if cv2.waitKey(1) == ord('q'):
            break

Thanks,
Jaka

spnsingh

jakaskerl

Thanks. Perhaps the point is there a good way to open the device, do some interactive things (across multiple cells), and then close then device.

Using with dai.Device(pipeline) as device everytime is slow because opening and closing the device takes 2-3 seconds each.

spnsingh

Otherwise, yes, I agree that this approach does resolve the connection issue. Thank you!

jakaskerl

Hi @spnsingh
The issue is that by only booting the device and not creating queues, the device has nowhere to put the frames so you have to specify the nodes to non-blocking.

Thanks
Jaka