OAK-D-PRO-POE-FF standalone mode

Irena

Hi!

I've recently started using the OAK-D-PRO-POE-FF and I have some doubts regarding its standalone mode operation.

I've tried the example code provided on GitHub, which can be found at the following link, gen2-cumulative-object-counting/standalone , but I have a few questions.

Once the pipeline is flashed onto the device, my intention is for the code (pipeline) to automatically launch as soon as the camera starts.

When run the oak.py, flashing pipeline, it's just show the progrees task Flashing progress: 99.7% After that I've checked bootloader_version.py and seems that the pipeline was flashed.

However, I noticed that when I disconnect and reconnect the device, the pipeline doesn't start running automatically. In other words, it seems that it requires a code launched from the host in order to start executing, just like in the provided example. That’s ok? Is this the expected behavior?

There are somebady can provide me with some tips on achieving a standalone mode where the pipeline starts as soon as the device is switched on?

Thank you in advance!

jakaskerl

Hi Irena
The pipeline should initialize as soon as you plug the power in the device with the flashed application.
One thing worth noting is that this is doesn't happen instantly. First the device has to be initialized, then the application. This takes some time (rougly 30s if i recall correctly).
Can you confirm this is the case with your device as well?

Thanks,
Jaka

Irena

Hi jakaskerl !

In my casi there's no conexion at all.

Let me try to explain that my host (PC) is waiting for a connection, and it is the device that should be able to connect to the host as soon as the device is switched on. However, this never happens. I'm not sure if this could be possible…

jakaskerl

Hi Irena
If you are using the unaltered version of cumulative-object-counting/standalone, then it's the device that's supposed to create the socket server, the host merely acts as a client.
This makes me think it's a problem with the flashing, since the pipeline should run automatically if flashed correctly.

Could you add the following code to the bottom of the oak.py script (modifying the flash code)?

(f, bl) = dai.DeviceBootloader.getFirstAvailableDevice()
bootloader = dai.DeviceBootloader(bl)
progress = lambda p : print(f'Flashing progress: {p*100:.1f}%')
(r, errmsg) = bootloader.flash(progress, pipeline)
if r: print("Flash OK")
else: print("Flash ERROR:", errmsg)

Thanks,
Jaka

Irena

Hi jakaskerl!

Here the result from your code:

To be precise, I have made slight modifications to the cumulative-object-counting algorithm. I have adapted the code to accommodate the functionality of switching between server and client roles. Therefore, the changes made are minimal in nature.

oak.py code changes looks like this:

/code above almost the same
script.setScript("""
import socket
import time
HOST_IP = "169.254.1.10"
sock = socket.socket()
sock.connect((HOST_IP, 5000))
/code below almost the same

host.py code changes looks like this:

/code above almost the same
server = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
server.bind(("0.0.0.0", 5000))
server.listen()
print("connect")
print("Waiting for connection")
connection, client = server.accept()
/code below almost the same

Thank you in advance!

jakaskerl

Hi Irena
Does the unchanged version work as it should in standalone mode?
Perhaps it is the switch of server and client that is causing the bug.

Thanks,
Jaka

Irena

Hi jakaskerl!

Upon my investigation, I can confirm that the unmodified code functions flawlessly, i.e. the camera acts as a server and the PC is the client that connects to it. I tested this by disconnecting and reconnecting the camera, and the host.py from script (gen2 cumulative object counting/standalone) works.

However, this setup does not align with the requirements I need to implement.

I tried a simple test code obtained from GitHub gen2-poe-tcp-streaming/poe-client (without any changes) to develop the opposite scenario. In this case, the camera is the client, and the host is the server working in standalone mode. After flashing the code onto the camera (oak.py), the process completed without any issues. Subsequently, I launched the server (host.py) on my PC, but I didn't get any response from the divece, looks like the pipeline dosen't work.

It is evident that there might be an important aspect missing from this process, but I haven't been able to identify it. I would appreciate your guidance.

Thank's a lot!

jakaskerl

Hi Irena
I was able to get it working by adding a try catch to the oak side, so it doesn't automatically break if no server is running.
Still took about a minute for the connection to occur though.

import depthai as dai
import time

pipeline = dai.Pipeline()

camRgb = pipeline.createColorCamera()
camRgb.setIspScale(2,3)

videoEnc = pipeline.create(dai.node.VideoEncoder)
videoEnc.setDefaultProfilePreset(30, dai.VideoEncoderProperties.Profile.MJPEG)
camRgb.video.link(videoEnc.input)

script = pipeline.create(dai.node.Script)
script.setProcessor(dai.ProcessorType.LEON_CSS)
videoEnc.bitstream.link(script.inputs['frame'])

script.setScript("""
# Enter your own IP!
HOST_IP = "192.168.2.117"

import socket
import time

while True:
    try:
        sock = socket.socket()
        sock.connect((HOST_IP, 8100))
        break
    except Exception as e:
        node.warn("Error connecting to {}:{}".format(HOST_IP, 8100))
        time.sleep(1)
    
while True:
    pck = node.io["frame"].get()
    data = pck.getData()
    ts = pck.getTimestamp()
    header = f"ABCDE " + str(ts.total_seconds()).ljust(18) + str(len(data)).ljust(8)
    # node.warn(f'>{header}<')
    sock.send(bytes(header, encoding='ascii'))
    sock.send(data)
""")

(f, bl) = dai.DeviceBootloader.getFirstAvailableDevice()
bootloader = dai.DeviceBootloader(bl)
progress = lambda p : print(f'Flashing progress: {p*100:.1f}%')
(r, errmsg) = bootloader.flash(progress, pipeline)
if r: print("Flash OK")
else: print("Flash ERROR:", errmsg)

Hope this helps,
Jaka

Irena

Hi jakaskerl!

This works.

Thank you very much!