camera adjuster fails

[VeeranjaneyuluToka]

Hi,

Nice and very detailed implementation.

I tried your stitched detailed notebook on our own set of images (no.of images are 3). but it fails with below error

---

error Traceback (most recent call last)
/tmp/ipykernel_2447557/2344847478.py in
8
9 cameras = camera_estimator.estimate(features, matches)
---> 10 cameras = camera_adjuster.adjust(features, matches, cameras)
11 cameras = wave_corrector.correct(cameras)

/mmdetection3d/tools/marine/stitching/camera_adjuster.py in adjust(self, features, pairwise_matches, estimated_cameras)
44
45 def adjust(self, features, pairwise_matches, estimated_cameras):
---> 46 b, cameras = self.adjuster.apply(features, pairwise_matches, estimated_cameras)
47 if not b:
48 raise StitchingError("Camera parameters adjusting failed.")

error: OpenCV(4.9.0) /io/opencv/modules/calib3d/src/compat_ptsetreg.cpp:125: error: (-215:Assertion failed) !err.empty() in function 'update'

I tried with all the camera adjusters, it works with no but none of other options works. I am not getting correct outuput with no adjuster. Could you please let me know what could be the reason and possible ways to analyse further on this issue.

Thanks,
Veeru.

[lukasalexanderweber]

Hey, how do you aquire your Images? Have a look on #135 (comment)

[lukasalexanderweber]

Oh I misread that you tried the original images from the notebook

[VeeranjaneyuluToka]

No, i am not trying with original images. I am trying with my own images which are generated with 3 cameras. The middle camera is positioned a bit higher than other two cameras.

[lukasalexanderweber]

https://stackoverflow.com/questions/78852689/slight-misalignment-issue-on-stitching-images-together-in-grid-like-manner/78855170#78855170

This setup will probably not work

[VeeranjaneyuluToka]

I could get stitched image with 3 images but the output looks like as follows

Detection: SIFT
Matcher: Homography
Adjuster: No
Estimator: Homography
wave_corrector: No
wrapper: Sperical
cropper=True
SeamFinder = dp_color
exposure_compensator = gain_blocks
blender=multiband

Do you think changing any of the above settings would give me better output? I also think as i am not applying bundleadjustment, this is how i am getting output. Let me know your comments?