luxonis
[BUG] Calibration Handler fails with Oak-1 OpenCV kit (no Camera data available)
https://github.com/luxonis/depthai-core/blob/main/include/depthai/device/CalibrationHandler.hpp https://github.com/luxonis/depthai-core/blob/main/src/device/CalibrationHandler.cpp The above Calibration handler can't handle the blank name returned by Original OpenCV AI Kit: Oak-1
This results in:
what(): There is no Camera data available corresponding to the the requested cameraId
As manifest here: https://github.com/luxonis/depthai-ros/pull/262 and here: https://github.com/luxonis/depthai-ros/issues/257#issuecomment-1480627256
[component_container-1] [INFO] [1679549436.807773594] [oak]: Device type:
Problem does not exist with OAK-1-MAX
Can I know when this OAK-1 eas purchased ? the initial versions didn't come with Eeprom to store the calibration data for them.
@saching13 it was an open box item on Ebay. https://www.ebay.com/itm/295571240836
Side of the box barcode says: "A00109", box is labeled as "AI KIT".
Regardless of when it was purchased, how can we work around this nuance?
ok. can you try this and see what error it gives ? If it throws an error. the work around would be to do an external calibration and use those intrinsics by saving them on host.
@saching13 can you explain the workaround in context of the referenced problems, aka simply starting the ROS nodes to publish any topic related to the cameras.
lol I now see there was a more complete example at: https://github.com/luxonis/depthai-python/blob/main/examples/calibration/calibration_reader.py
It returned this error:
ubuntu@ubuntu:~$ python3 calibration_reader.py Stored.json
Traceback (most recent call last):
File "calibration_reader.py", line 17, in <module>
M_rgb, width, height = calibData.getDefaultIntrinsics(dai.CameraBoardSocket.RGB)
RuntimeError: There is no Camera data available corresponding to the the requested cameraId
Here are the contents of the attempt to store the config to a file...
ubuntu@ubuntu:~$ cat Stored.json
{
"batchName": "",
"batchTime": 0,
"boardConf": "",
"boardCustom": "",
"boardName": "",
"boardOptions": 0,
"boardRev": "",
"cameraData": [],
"hardwareConf": "",
"imuExtrinsics": {
"rotationMatrix": [],
"specTranslation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
},
"toCameraSocket": -1,
"translation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
}
},
"miscellaneousData": [],
"productName": "",
"stereoRectificationData": {
"leftCameraSocket": -1,
"rectifiedRotationLeft": [],
"rectifiedRotationRight": [],
"rightCameraSocket": -1
},
"version": 7
}
For posterity, here is my OAK-1-MAX
ubuntu@ubuntu:~$ python3 calibration_reader.py Stored.json
RGB Camera Default intrinsics...
[[2986.73291015625, 0.0, 2004.2607421875], [0.0, 2986.73291015625, 1510.3992919921875], [0.0, 0.0, 1.0]]
4000
3000
RGB Camera Default intrinsics...
[[2986.73291015625, 0.0, 2004.2607421875], [0.0, 2986.73291015625, 1510.3992919921875], [0.0, 0.0, 1.0]]
4000
3000
RGB Camera resized intrinsics... 3840 x 2160
[[2.86726343e+03 0.00000000e+00 1.92409021e+03]
[0.00000000e+00 2.86726343e+03 1.08998328e+03]
[0.00000000e+00 0.00000000e+00 1.00000000e+00]]
RGB Camera resized intrinsics... 4056 x 3040
[[3.02854736e+03 0.00000000e+00 2.03232056e+03]
[0.00000000e+00 3.02854736e+03 1.53054480e+03]
[0.00000000e+00 0.00000000e+00 1.00000000e+00]]
Traceback (most recent call last):
File "calibration_reader.py", line 53, in <module>
M_left, width, height = calibData.getDefaultIntrinsics(dai.CameraBoardSocket.LEFT)
RuntimeError: There is no Camera data available corresponding to the the requested cameraId
and its JSON
ubuntu@ubuntu:~$ cat Stored.json
{
"batchName": "Maxwell",
"batchTime": 1667974639,
"boardConf": "nIR-C02M00-00",
"boardCustom": "",
"boardName": "NG9093",
"boardOptions": 0,
"boardRev": "R3M0E4",
"cameraData": [
[
0,
{
"cameraType": 0,
"distortionCoeff": [
-7.164836883544922,
20.304182052612305,
0.00017078577366191894,
-0.00011530541814863682,
-16.3474063873291,
-7.179931163787842,
20.356590270996094,
-16.410999298095703,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"extrinsics": {
"rotationMatrix": [
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
]
],
"specTranslation": {
"x": -0.0,
"y": -0.0,
"z": -0.0
},
"toCameraSocket": -1,
"translation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
}
},
"height": 3000,
"intrinsicMatrix": [
[
2986.73291015625,
0.0,
2004.2607421875
],
[
0.0,
2986.73291015625,
1510.3992919921875
],
[
0.0,
0.0,
1.0
]
],
"lensPosition": 130,
"specHfovDeg": 68.7938003540039,
"width": 4000
}
]
],
"hardwareConf": "F0-FV00-BC000",
"imuExtrinsics": {
"rotationMatrix": [
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
]
],
"specTranslation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
},
"toCameraSocket": -1,
"translation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
}
},
"miscellaneousData": [],
"productName": "OAK-1-MAX",
"stereoRectificationData": {
"leftCameraSocket": 255,
"rectifiedRotationLeft": [
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
]
],
"rectifiedRotationRight": [
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
],
[
0.0,
0.0,
0.0
]
],
"rightCameraSocket": 255
},
"version": 7
}
@saching13 can you explain the workaround in context of the referenced problems, aka simply starting the ROS nodes to publish any topic related to the cameras.
you can calibrate and save the json file on the host and pass it as a param to it. Adam can help on how to set that up.
@saching13 yeah, my goal would be to simply make it work transparently for anyone that happens to own said camera variant, and chooses to use the existing ROS nodes. In the mean time I simply bought a new camera.
my goal would be to simply make it work transparently for anyone that happens to own said camera variant, and chooses to use the existing ROS nodes.
I understand the concern. But since it is missing what we can do is override and publish a standard dummy camera info for that as default with a warning. Unless the user can calibrate and add an argument of that.
sounds completely reasonable! This is my first week with you all's code base so, forgive some of my uninformed questions. I appreciate the help.
@Serafadam I'm sadly stuck on Noetic on a few of my instances, will this be backported to the other branches as well?
Hi @MAVProxyUser , sorry, this got lost among notifications, yes it was backported soon afterwards to Noetic.