Availability of datasets used in the paper

[mdouskos]

Hi @puzzlepaint, besides the dataset provided in #16, are there any plans to make available the datasets used in the paper, i.e. the ones corresponding to the cameras mentioned in Table 2 of the arXiv version? Thank you in advance!

[puzzlepaint]

Generally speaking, since I finished my PhD work roughly two years ago, I currently have no plans for this project aside from doing minimal maintenance (in my spare time). The datasets weren't originally intended for release and might require some checking before upload, for example there are other persons visible on the GoPro images due to its large field-of-view, so it would be necessary to mask them out. But if you are interested in a specific dataset, I would probably be able to upload it. Table 2 in the arXiv version only lists the cameras used. Are you interested in the datasets for a specific experiment, for example Figure 11?

[spdfghi]

Hi, I'm impressed by your great work, and now I want to write a paper with citing your dataset D435I. I run the code you provide but did not get same result as your paper. For Central Generic case, I got reprojection error around 0.055, there is a large bias against 0.023 in paper. I'm sure I set the right parameter, now I guess some reasons:

1. There are some bad data was not used in paper, since the image number in Is it possible to provide some open datasets for test？ #16 is 1379, while in paper is 1347.
2. I used the cuda version.
3. Not all images was used to calibration, since we saw there are train and test partition in paper.
   Can you give me some advice?
   @puzzlepaint