MAFAT Satellite Vision Challenge Forum

Go back to competition Back to thread list Post in this thread

> Relative Azimuth and Resolution

So we are given a horizonal Azimuth, is this relative to the center of the scene being viewed?

We are provided with a pixel-meter Resolution value, but this should differ in the X and Y axes. Is it an average of them? Is it estimated from the intrinsic parameters of the camera or the known size of the scene/satellite angle? I could make useful estimates of all of these things, but this would require me to assume that the same satellite is used for every scene.

Posted by: ssleder @ April 18, 2023, 7:55 p.m.

To add on, if the Resolution would be inaccurate from the origin due to lens distortion, curvature of the planet, and the satellite angle again.

Posted by: ssleder @ April 18, 2023, 8:08 p.m.

The Resolution\Azimuth - is the average resolution\azimuth of the original full image.
The original image is divided into 1280x1280 size frames (the competition dataset) so you might expect some drift from the declared resolution\azimuth to the one in practice.
You may assume camera intrinsic are constant across the entire dataset.
MAFAT Challenge Team

Posted by: MAFAT_Challenge @ April 20, 2023, 11:45 a.m.

I can do a lot if I can assume the camera intrinsic are the same. Is it safe to make this assumption for the testing data as well?

Posted by: ssleder @ April 21, 2023, 1:50 a.m.

You can make this assumption on the test data as well.
MAFAT Challenge Team

Posted by: MAFAT_Challenge @ April 23, 2023, 7 a.m.
Post in this thread