dpizzle's profile - activity

It looks like you are expecting rvec to be euler angles in radians. You need to convert rvec to a matrix first using cv

It looks like you are expecting rvec to be euler angles in radians. Take a look at: https://stackoverflow.com/questions

I don't have any definitive proof, but from past experience I'm thinking the bigger the marker would be better than many

Maybe try dialating before houghlines? https://docs.opencv.org/2.4/doc/tutorials/imgproc/erosion_dilatation/erosion_dil

Here's what I would try, crop a larger area. Run hough line. Make two lists of horizontal lines and vertical lines by

I think some front facing cameras will flip the image. Make sure that is not happening. https://www.quora.com/Why-does

Not quite sure of the problem, obviously make sure you calibration, etc is correct. Here's another way of tackling. Ass

Not sure exactly what you need. Those 4 sets of numbers are the 2D coordinates of the four corners of the marker in the

I don't think Aruco can return the "quality of fit". Maybe look into applying a Kalman filter that will smooth the data

Have you looked at PCL (point cloud library) yet? I might be wrong but not sure opencv can do this. Also do a search for Mat to PCL for code examples of how to integrate opencv with PCL.

The region capture plugin is gonna be specific to Unity and Vuforia, won't be useful at all to opencv. But this is how a lot of the commercial apps do AR coloring books. https://github.com/maximrouf/RegionCa...

I've done an implementation of this in Unity with Vuforia and Region Capture (search github). But here would be the steps in opencv.

1. Detect the marker which is the B/W coloring page (something like sturkmen's answer might work)
2. From detecting the marker you know the coordinates of where in the camera stream the page is
3. Using the four corners, cutout and unwarp the colored page from the camera stream so it fills a square
4. Use this square image as a the texture for your UV map on your 3d model
5. Display 3d model on marker

The key is the UV map of the model has to somewhat match the actual coloring page. You'll have to do some tricks to get the backside of the model, like doubling up with the front side of the UV map in that area. Search UV map on google images to get an idea of what they are if you don't know. Good luck.

Smaller grid markers will work better at extreme angles and larger distances. Depends on your set up whether this will make a big difference or not.

I think the contrib module is based on Aruco 1.0. And that website is the new 2.0 version which offers many improvements. The old Aruco had what were called boards (known as marker maps in 2.0) but were planar only. Try to go back to the previous code you were using if you need that feature.

Here's some more discussions of the same issue: https://github.com/chili-epfl/chilitags/issues/19 (https://github.com/chili-epfl/chilita...) https://github.com/chili-epfl/chilitags/issues/82 (https://github.com/chili-epfl/chilita...)

Are you sure the markers are totally flat? Can't quite tell but the paper looks a little curved.

Does the "flipping" happen more when marker is at the edges of the frame (which I think means it's distortion related, but not sure, im still new to cv)

Try turning on corner refinement and messing with cornerRefinementWinSize? SolvePnP is only as good as the data it has to work.

SolvePnP allows you to pass in rvec and tvec as initial guesses. Maybe you could check the angle to the camera and pass it in an initial angle somehow.

I recently switched to apriltags (still using opencv's Solve PnP for pose) and am getting much better results, but it's a little slow.