Hi, I find that Jetpack 6.1 includes VPI 3.2 having a update with AprilTag Detector.
As we know, for example, a 36h11 tag contains 10 x 10 pixels, including an outer border of all white pixels and an inner border of all black pixels (which means actually tag size is 8x8).
By using the example code in document, I could draw corner point at 8x8.
*** Please note that this reply is generated by LLM automatically ***
I’d be happy to help you with your question about the AprilTag Detector in VPI 3.2, which is included in Jetpack 6.1.
Firstly, I’d like to clarify that the AprilTag Detector is a feature in VPI (Vision Programming Interface) that allows for the detection of AprilTags, which are a type of visual marker used for various applications such as robotics, augmented reality, and computer vision.
Regarding your issue, it seems that you’re experiencing difficulties with the detection of AprilTags when the outer border of the tag is black. You’ve tried using the example code provided in the documentation, but you’re unable to find a parameter to set the outer border color.
After reviewing the available information, I found that the AprilTag Detector in VPI 3.2 uses a default set of parameters for detecting AprilTags, which may not be configurable through a specific parameter for outer border color.
However, I can suggest a few potential solutions to help you overcome this issue:
Check the documentation: I recommend reviewing the VPI 3.2 documentation again to see if there are any mentions of configurable parameters for the AprilTag Detector that might help you achieve your goal.
Use a different detection algorithm: Depending on your specific use case, you might consider using a different detection algorithm that is more robust to variations in the tag’s appearance, such as the ArUco marker detection algorithm.
Pre-processing of the image: You could try applying pre-processing techniques to the image before passing it to the AprilTag Detector, such as converting the image to grayscale or applying a threshold to enhance the contrast between the tag and the background.
Custom implementation: If none of the above solutions work for you, you might need to consider implementing a custom solution using a computer vision library such as OpenCV, which would allow you to have more control over the detection process.
In comparison to other computer vision libraries, VPI is a relatively high-level library that provides a simple and easy-to-use API for various computer vision tasks, including object detection, tracking, and image processing. While it may not offer the same level of customization as lower-level libraries like OpenCV, it can still be a powerful tool for many applications.
I hope this helps you find a solution to your issue! If you have any further questions or need more specific guidance, feel free to ask.
Thanks for the reply.
I think I have set the correct VPIAprilTagFamily.
But as I said, some calibration board do not use the default family pattern.
They may change the border size or outer border color.
(For example, the picture in last post has a 2px black border.)
In this case, VPI AprilTag Detector could not detect correctly.
I think nowadays, apriltag calibration board with 2 bit black borders becomes common, which makes the corner squares between the tags have more accurate localization of the tag corners.
We can find many calibration boards with 2 bit black border on the net shop.
And some apriltag detection library provides parameter to set border width.
For example, ETH Zürich’s tag detector has a blackBorder like:
Anyway, I will try to find a ‘normal’ calibration board.
Thanks for the reply again.
