Calibration error

PetrNovota · Dec 12, 2022

@"JamesYeh" I would like you to try one little experiment.
in your depthai/ folder, locate the calibrate.py file. open it with any text editor or an IDE like Pycharm.
On lines 611-614
you can locate this code:

if self.args.factoryCalibration:
    self.device.flashFactoryCalibration(calibration_handler)
is_write_succesful = self.device.flashCalibration(calibration_handler)

change it to

if self.args.factoryCalibration:
    self.device.flashFactoryCalibration(calibration_handler)
is_write_succesful = self.device.flashCalibration2(calibration_handler)

The change is int the self.device.flashCalibration2

It can happen that this code piece is not on above mentioned lines, it depends on the depthai branch. Either way yor calibrate.py file should contain those lines so you can find them.

When it is done, run the calibration again, the same way as before. Now, it should end with en Error printed out in the terminal. Send us the error please.

JamesYeh · Dec 12, 2022

It's interesting that I don't see the codes you sent me. Here is what I installed for the calibration process in your website https://docs.luxonis.com/en/latest/pages/calibration/. Where did you get the calibrate.py?

git clone https://github.com/luxonis/depthai.git
cd depthai
git checkout lite_calibration

JamesYeh · Dec 12, 2022

Here are the codes at the end of the calibrate.py

resImage = None
if not self.device.isClosed():
dev_info = self.device.getDeviceInfo()
mx_serial_id = dev_info.getMxId()
calib_dest_path = dest_path + '/' + mx_serial_id + '.json'
calibration_handler.eepromToJsonFile(calib_dest_path)
is_write_succesful = False

            try:
                is_write_succesful = self.device.flashCalibration(
                    calibration_handler)
            except:
                print("Writing in except...")
                is_write_succesful = self.device.flashCalibration(
                    calibration_handler)
            if is_write_succesful:
                resImage = create_blank(900, 512, rgb_color=green)
                text = "Calibration Succesful with"
                cv2.putText(resImage, text, (10, 250),
                            font, 2, (0, 0, 0), 2)
                text = "Epipolar error of " + str(epiploar_error)
                cv2.putText(resImage, text, (10, 300),
                            font, 2, (0, 0, 0), 2)
            else:
                resImage = create_blank(900, 512, rgb_color=red)
                text = "EEprom Write Failed!! " + str(epiploar_error)
                cv2.putText(resImage, text, (10, 250),
                            font, 2, (0, 0, 0), 2)
                text = "Try recalibrating !!"
                cv2.putText(resImage, text, (10, 300),
                            font, 2, (0, 0, 0), 2)
        else:
            calib_dest_path = dest_path + '/depthai_calib.json'
            # calibration_handler.eepromToJsonFile(calib_dest_path)
            resImage = create_blank(900, 512, rgb_color=red)
            text = "Calibratin succesful. " + str(epiploar_error)
            cv2.putText(resImage, text, (10, 250), font, 2, (0, 0, 0), 2)
            # text = "Device not found to write to EEPROM"
            # cv2.putText(resImage, text, (10, 300), font, 2, (0, 0, 0), 2)

        if resImage is not None:
            cv2.imshow("Result Image", resImage)
            cv2.waitKey(0)
    except AssertionError as e:
        print("[ERROR] " + str(e))
        raise SystemExit(1)

def run(self):
    if 'capture' in self.args.mode:
        try:
            if Path('dataset').exists():
                shutil.rmtree('dataset/')
            Path("dataset/left").mkdir(parents=True, exist_ok=True)
            Path("dataset/right").mkdir(parents=True, exist_ok=True)
            if not self.args.disableRgb:
                Path("dataset/rgb").mkdir(parents=True, exist_ok=True)
        except OSError:
            traceback.print_exc()
            print("An error occurred trying to create image dataset directories!")
            raise SystemExit(1)
        self.show_info_frame()
        self.capture_images()
    self.dataset_path = str(Path("dataset").absolute())
    if 'process' in self.args.mode:
        self.calibrate()
    print('py: DONE.')

if name == "main":
Main().run()

PetrNovota · Dec 12, 2022

JamesYeh in the branch lite_calibration that you are using it is in
depthai/calibrate.py
in there, lines 570-576 change them from:

try:
    is_write_succesful = self.device.flashCalibration(calibration_handler)
except:
    print("Writing in except...")
    is_write_succesful = self.device.flashCalibration(calibration_handler)

to

self.device.flashCalibration2(calibration_handler)

so it will now look like this:

...
...
calib_dest_path = dest_path + '/' + mx_serial_id + '.json'
calibration_handler.eepromToJsonFile(calib_dest_path)
is_write_succesful = False

is_write_succesful = self.device.flashCalibration2(calibration_handler)
if is_write_succesful:
    resImage = create_blank(900, 512, rgb_color=green)
    ...
    ...
    ...

and run calibration again, you should get an Error, send it to us

JamesYeh · Dec 12, 2022

I changed the calibrate.py and ran it again. It is still the same. See the screen shots...

PetrNovota · Dec 12, 2022

JamesYeh thank you, good job. Now, I want to ask you to do one more thing (consists of two sub tasks ).

With editor of your choice create a new file called get_calibration.py and save it to your depthai folder, so it will look like this: dephtai/get_calibration.py
The content of the get_calibration.py file is:

import depthai as dai

with dai.Device() as device:
    print(device.readCalibration2().eepromToJson())

and in depthai folder run python get_calibration.py
This will print the current device calibration in your terminal. Please copy it and send it to us.

Second task is the same one you have done before. In depthai/resources there is your <device_mx_id(long number).json> file. Please send its contents to us as well.

Thank you for your patience.
Best regards
Petr

JamesYeh · Dec 12, 2022

Ok will try it. And, is it possible to run the test again with the sames images in Dataset folder under depthai folder? It would be more consistent for this debug with the same image set. Thanks!

JamesYeh · Dec 12, 2022

here is the .json text file.

{
"batchName": "",
"batchTime": 0,
"boardConf": "",
"boardCustom": "",
"boardName": "OAK-D-LITE",
"boardOptions": 0,
"boardRev": "R0M0E0",
"cameraData": [
[
1,
{
"cameraType": 0,
"distortionCoeff": [
-28.19111442565918,
319.1015625,
0.0021235786844044924,
-0.00796385295689106,
-973.7730712890625,
-28.19227409362793,
319.0585021972656,
-973.5380249023438,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"extrinsics": {
"rotationMatrix": [
[
0.9998503923416138,
-0.0063782683573663235,
-0.016076868399977684
],
[
0.0063193743117153645,
0.9999731183052063,
-0.003711411962285638
],
[
0.01610010862350464,
0.003609261242672801,
0.999863862991333
]
],
"specTranslation": {
"x": -7.5,
"y": 0.0,
"z": 0.0
},
"toCameraSocket": 2,
"translation": {
"x": -7.544513702392578,
"y": -0.06922637671232224,
"z": -0.03127232566475868
}
},
"height": 480,
"intrinsicMatrix": [
[
455.9234619140625,
0.0,
319.5633239746094
],
[
0.0,
456.3455505371094,
230.58319091796875
],
[
0.0,
0.0,
1.0
]
],
"lensPosition": 0,
"specHfovDeg": 72.9000015258789,
"width": 640
}
],
[
2,
{
"cameraType": 0,
"distortionCoeff": [
21.66294288635254,
-312.0931396484375,
0.0010793297551572323,
-0.0007400921895168722,
906.7825927734375,
21.852170944213867,
-314.07861328125,
912.0392456054688,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"extrinsics": {
"rotationMatrix": [
[
0.9998428821563721,
0.00951849389821291,
-0.014950711280107498
],
[
-0.009514091536402702,
0.9999546408653259,
0.00036551340599544346
],
[
0.014953513629734516,
-0.00022321353026200086,
0.9998881220817566
]
],
"specTranslation": {
"x": 3.75,
"y": 0.0,
"z": 0.0
},
"toCameraSocket": 0,
"translation": {
"x": 3.9650027751922607,
"y": 0.006998621392995119,
"z": -0.5213539600372314
}
},
"height": 480,
"intrinsicMatrix": [
[
455.4236755371094,
0.0,
304.4522399902344
],
[
0.0,
455.9582824707031,
240.3455352783203
],
[
0.0,
0.0,
1.0
]
],
"lensPosition": 0,
"specHfovDeg": 72.9000015258789,
"width": 640
}
],
[
0,
{
"cameraType": 0,
"distortionCoeff": [
30.001792907714844,
-27.810789108276367,
0.0005656384746544063,
0.0028413927648216486,
170.4805908203125,
30.07124900817871,
-31.029281616210938,
174.8137664794922,
0.0,
0.0,
0.0,
0.0,
0.0,
0.0
],
"extrinsics": {
"rotationMatrix": [],
"specTranslation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
},
"toCameraSocket": -1,
"translation": {
"x": 0.0,
"y": 0.0,
"z": 0.0
}
},
"height": 1080,
"intrinsicMatrix": [
[
1493.0745849609375,
0.0,
1028.433837890625
],
[
0.0,
1489.53271484375,
517.1607055664063
],
[
0.0,
0.0,
1.0
]
],
"lensPosition": 135,
"specHfovDeg": 68.7938003540039,
"width": 1920
}
]
],
"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": [
[
0.9999244809150696,
0.0028120274655520916,
-0.011965838260948658
],
[
-0.002833560574799776,
0.9999943971633911,
-0.001782975741662085
],
[
0.01196075789630413,
0.001816747011616826,
0.9999268054962158
]
],
"rectifiedRotationRight": [
[
0.9999493360519409,
0.009175259619951248,
0.00414483156055212
],
[
-0.009182704612612724,
0.9999562501907349,
0.0017808872507885098
],
[
-0.004128310363739729,
-0.0018188577378168702,
0.9999898076057434
]
],
"rightCameraSocket": 2
},
"version": 7
}

JamesYeh · Dec 12, 2022

From the EEprom data:
PS C:\GG\OAK-Calibrate\OAK-Calibrate\depthai> python get_calibrate.py
{'batchName': '', 'batchTime': 0, 'boardConf': '', 'boardCustom': '', 'boardName': 'OAK-D-LITE', 'boardOptions': 0, 'boardRev': 'R1M1E3', 'cameraData': [[0, {'cameraType': 0, 'distortionCoeff': [-4.465783596038818, 9.97364616394043, -0.00019955226161982864, -0.0010946181137114763, -4.277606964111328, -4.544586181640625, 10.269360542297363, -4.632837295532227, 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': 2160, 'intrinsicMatrix': [[2969.224609375, 0.0, 2027.668701171875], [0.0, 2969.224609375, 1030.9114990234375], [0.0, 0.0, 1.0]], 'lensPosition': 0, 'specHfovDeg': 68.7938003540039, 'width': 3840}], [1, {'cameraType': 0, 'distortionCoeff': [-5.925610542297363, 13.577345848083496, 0.0005030203028582036, -0.0037203540559858084, 2.8491311073303223, -5.928144454956055, 13.576969146728516, 2.8172836303710938, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0], 'extrinsics': {'rotationMatrix': [[0.9999157190322876, -0.005482868291437626, -0.011770736426115036], [0.005498494952917099, 0.9999840259552002, 0.0012956152204424143], [0.011763445101678371, -0.0013602273538708687, 0.9999299049377441]], 'specTranslation': {'x': -7.5, 'y': 0.0, 'z': 0.0}, 'toCameraSocket': 2, 'translation': {'x': -7.459471702575684, 'y': -0.11444893479347229, 'z': -0.08647604286670685}}, 'height': 480, 'intrinsicMatrix': [[456.83233642578125, 0.0, 322.490234375], [0.0, 456.83233642578125, 228.93768310546875], [0.0, 0.0, 1.0]], 'lensPosition': 0, 'specHfovDeg': 72.9000015258789, 'width': 640}], [2, {'cameraType': 0, 'distortionCoeff': [15.70577621459961, -133.31661987304688, -0.0004125449631828815, -0.002012420678511262, 418.70599365234375, 15.833054542541504, -134.20359802246094, 420.8946533203125, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0], 'extrinsics': {'rotationMatrix': [[0.9999483227729797, 0.008442151360213757, -0.005667971912771463], [-0.008480061776936054, 0.9999415874481201, -0.0066981068812310696], [0.005611094646155834, 0.00674582552164793, 0.9999614953994751]], 'specTranslation': {'x': 3.75, 'y': 0.0, 'z': 0.0}, 'toCameraSocket': 0, 'translation': {'x': 3.7240123748779297, 'y': 0.07089735567569733, 'z': -0.19739195704460144}}, 'height': 480, 'intrinsicMatrix': [[455.19598388671875, 0.0, 302.0895690917969], [0.0, 455.19598388671875, 237.84361267089844], [0.0, 0.0, 1.0]], 'lensPosition': 0, 'specHfovDeg': 72.9000015258789, 'width': 640}]], 'hardwareConf': '', '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': '', 'stereoRectificationData': {'leftCameraSocket': 1, 'rectifiedRotationLeft': [[0.9999515414237976, 0.009842065162956715, -0.00015885815082583576], [-0.009841952472925186, 0.9999513030052185, 0.0006939409067854285], [0.0001656802196521312, -0.0006923438049852848, 0.9999997615814209]], 'rectifiedRotationRight': [[0.9998151659965515, 0.015339930541813374, 0.011590640991926193], [-0.015331892296671867, 0.999882161617279, -0.0007820441969670355], [-0.011601271107792854, 0.0006041931919753551, 0.9999325275421143]], 'rightCameraSocket': 2}, 'version': 6}
PS C:\GG\OAK-Calibrate\OAK-Calibrate\depthai>

PetrNovota · Dec 12, 2022

JamesYeh Thank you very much. We know what the problem is, we can solve it, we just have to discuss what the best possible course of action is. Ill get back to you as soon as I can.

JamesYeh · Dec 12, 2022

Thanks! Are we talking about 1-2 days or more? We've tried the calibration in the past week without success...

PetrNovota · Dec 12, 2022

JamesYeh Hello James. The fix is done.
Please pull latest changes from lite_calibration branch, i.e. git fetch and git pull.
After that you can do the calibration one more time, this time it should succeed. If not, let me know, but it is very unlikely.

JamesYeh · Dec 12, 2022

can you please elaborate more how to proceed with some details? Where is the lite_calibration branch, etc... Thanks!

PetrNovota · Dec 12, 2022

@JamesYeh if you dont want to repeat the image capturing, because you already have those, then you can comment out lines 611-623 in calibrate.py script, i.e. add # in front of each line.

PetrNovota · Dec 12, 2022

JamesYeh You wrote:
git clone https://github.com/luxonis/depthai.git
cd depthai
git checkout lite_calibration

Right now, open your terminal and cd into your deptahi dir. When you are in the depthai dir, you enter commands
git fetch confirm with enter
git pull confirm with enter

PetrNovota · Dec 12, 2022

@JamesYeh then you can run python calibrate.py again, I suggest you comment out lines I mentioned, then capturing images will be skipped. You can do this if you still have captured images from before in deptahi/dataset/ folder

JamesYeh · Dec 12, 2022

after fetch and pull, I got this error message

JamesYeh · Dec 12, 2022

Then I ran the calibration again but it is still the same error message with EEprom Write failed...

PetrNovota · Dec 12, 2022

JamesYeh I am sorry didnt realize we made changes.
Use git restore first, then git fetch then git pull.
Only when its successful, run calibration again

JamesYeh · Dec 12, 2022

WE MADE IT!!!