About function get_rotation_angle
Closed this issue · 2 comments
Hello, @StevenBanama . When I'm trying to preprocess some new dataset, I met some problems about function get_rotation_angle where giving out assertion error on cv2.solvePnP. I check cv2 document about cv2.solvePnP and find something as follows.
If you are using Python:
- Numpy array slices won't work as input because solvePnP requires contiguous arrays (enforced by the assertion using cv::Mat::checkVector() around line 55 of modules/calib3d/src/solvepnp.cpp version 2.4.9)
- The P3P algorithm requires image points to be in an array of shape (N,1,2) due to its calling of cv::undistortPoints (around line 75 of modules/calib3d/src/solvepnp.cpp version 2.4.9) which requires 2-channel information.
- Thus, given some data D = np.array(...) where D.shape = (N,M), in order to use a subset of it as, e.g., imagePoints, one must effectively copy it into a new array: imagePoints = np.ascontiguousarray(D[:,:2]).reshape((N,1,2))
I think that adding
landmarks = np.ascontiguousarray(landmarks.reshape((landmarks.shape[0], 1, landmarks.shape[1])))
after landmarks would be helpful.
Hello, @StevenBanama . When I'm trying to preprocess some new dataset, I met some problems about function get_rotation_angle where giving out assertion error on cv2.solvePnP. I check cv2 document about cv2.solvePnP and find something as follows.
If you are using Python:
- Numpy array slices won't work as input because solvePnP requires contiguous arrays (enforced by the assertion using cv::Mat::checkVector() around line 55 of modules/calib3d/src/solvepnp.cpp version 2.4.9)
- The P3P algorithm requires image points to be in an array of shape (N,1,2) due to its calling of cv::undistortPoints (around line 75 of modules/calib3d/src/solvepnp.cpp version 2.4.9) which requires 2-channel information.
- Thus, given some data D = np.array(...) where D.shape = (N,M), in order to use a subset of it as, e.g., imagePoints, one must effectively copy it into a new array: imagePoints = np.ascontiguousarray(D[:,:2]).reshape((N,1,2))
I think that adding
landmarks = np.ascontiguousarray(landmarks.reshape((landmarks.shape[0], 1, landmarks.shape[1])))
after landmarks would be helpful.
BTW, I have tried this method and successfully run the code. Thank you for your implementation!
sorry, lately reply. "np.ascontiguousarray() or np.copy()" may meet your demands. Meanwhile you can proccess you dataset with landmarks when you garante it without side-face.