Hi We would like to check the z, x, y when the camera looks downwards. But there is some issue. Please help us to handle this issue. Here is the function of our development with your PoE model. 1. Activate the system for 1 minute every 2 minutes 2. Save image: Save only the left image 3. Extract x, y, depth, z values for points of interest (Point_base.dat, Point_wl.dat) and save as text Issue we met now. 1. When the camera is looking straight ahead, x, y, depth values are accurate. no problem. 2. When the camera is looking diagonally downward, depth(z) is accurate, but x, y errors occur 3. In my personal opinion, I think that calibration information (left_camera_matrix, right_camera_matrix, rotation_matrix, translation_vector) should be utilized 4. However, it is not clear whether the calibration information in this code is utilized, and it is understood that the calibration information is inside the camera So i think that I need your help to get correct x, y information even though the camera is looking downward direction. Here also i added our code so far. For your information. [source code ](https://drive.google.com/drive/folders/1kIR5yAvRnhfzZKAqydyw4TGaIYhlQKqM?usp=drive_link "source code so far to get correct x, y, z value") Best regards, Ryan.

Hi @"RyanLee"#p26941 Yes, I mean the IMU rotation vector. Thanks, Jaka

When the camera looks downwards, x and y value is not correct

RyanLee

We would like to check the z, x, y when the camera looks downwards. But there is some issue. Please help us to handle this issue.

Here is the function of our development with your PoE model.

Activate the system for 1 minute every 2 minutes
Save image: Save only the left image
Extract x, y, depth, z values for points of interest (Point_base.dat, Point_wl.dat) and save as text

Issue we met now.

When the camera is looking straight ahead, x, y, depth values are accurate. no problem.
When the camera is looking diagonally downward, depth(z) is accurate, but x, y errors occur
In my personal opinion, I think that calibration information (left_camera_matrix, right_camera_matrix, rotation_matrix, translation_vector) should be utilized
However, it is not clear whether the calibration information in this code is utilized, and it is understood that the calibration information is inside the camera

So i think that I need your help to get correct x, y information even though the camera is looking downward direction.

Here also i added our code so far. For your information.

source code

Best regards,

Ryan.

jakaskerl

RyanLee
If you wish to view X, Y and Z from the world coordinate frame, you need to account for the transform from the world to camera reference frames. You can do this via rotation vector. This has to be done manually (on host), the device only calculates the position from its own camera's reference frame.

Thanks,
Jaka

RyanLee

Hi,

I have a question you mean that "IMU Rotation Vector" ?? if not could you explain more about the rotation vector?

https://docs.luxonis.com/software/depthai/examples/imu_rotation_vector?searchTokens=%5B%22rotation%22%2C%22vector%22%2C%22Rotation%22%2C%22Vector%22%5D

Best regards,

Ryan.

jakaskerl

Hi RyanLee
Yes, I mean the IMU rotation vector.

Thanks,
Jaka

RyanLee

Thanks jakaskerl