While going through the test, training and valid files, i noticed many landmarks have 0 for the x,y coordinates. This is not mentioned in the paper - is it an error in uploading? Attached below is a screenshot.
Would you be able to send the correct files on veer@artimed.in if this is the issue? Thank you.
Also, by any chance is the resolution of the images compromised during the download process. Would you be able to provide us with the link to the original files.
@veerparikh27 Thank you for your interest in our work. Let me address your questions:
Regarding the zero coordinates: This is not an error. In our first phase, we focused on verifying the accuracy of existing landmark annotations rather than adding missing points. You can use the landmarks_v variable during inference to check whether a specific landmark is annotated.
About the image resolution: The images in the dataset are in their original resolution, which is slightly below 1k. We did not compress them to save disk space, so what you see is the native resolution of these X-ray images.
Hope this helps clarify your concerns. Please let us know if you have any other questions.
Hey @switch626, Thanks for the great work.
While going through the test, training and valid files, i noticed many landmarks have 0 for the x,y coordinates. This is not mentioned in the paper - is it an error in uploading? Attached below is a screenshot.
Would you be able to send the correct files on veer@artimed.in if this is the issue? Thank you.
Also, by any chance is the resolution of the images compromised during the download process. Would you be able to provide us with the link to the original files.