You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
You have a few incorrect points labelled here. I can tell just from looking at the numbers here that the line with "94" as the pixel x coordinate is missing a digit (should be 940-somthing), and you are missing several negatives in the y meters component in the last few points. The x values all seem plausible. This would explain why the distances we were seeing were fairly close, but the left-right alignment was very off.
Please try to redo this as soon as you have an opportunity. I would recommend starting with just 4 labelled points near the corners of the image and testing to see if the homography comes out correctly. Once we know those 4 are correct, we can add a few more points to help reduce any noise that exists in the labelling process.
I'd really like to get auto-pickup working before Nick gets to the meeting tonight so we can start right on autos when he gets there. Once the homography is fixed, we will need to deal with the double-detection issue we were seeing yesterday, but then we should have everything we need for auto-pickup to work as intended.
I'll try to get those new points done after school today. I also just realized I'm applying a camera transform added to the homography that is already transformed.
2024SwerveBase/src/main/java/frc/robot/subsystems/NoteDetection.java
Line 51 in fc0fdbf
You have a few incorrect points labelled here. I can tell just from looking at the numbers here that the line with "94" as the pixel x coordinate is missing a digit (should be 940-somthing), and you are missing several negatives in the y meters component in the last few points. The x values all seem plausible. This would explain why the distances we were seeing were fairly close, but the left-right alignment was very off.
Please try to redo this as soon as you have an opportunity. I would recommend starting with just 4 labelled points near the corners of the image and testing to see if the homography comes out correctly. Once we know those 4 are correct, we can add a few more points to help reduce any noise that exists in the labelling process.
I'd really like to get auto-pickup working before Nick gets to the meeting tonight so we can start right on autos when he gets there. Once the homography is fixed, we will need to deal with the double-detection issue we were seeing yesterday, but then we should have everything we need for auto-pickup to work as intended.
@Alex-idk
The text was updated successfully, but these errors were encountered: